Namespace(accumulate=1, batch_norm=False, batch_size=8, clip_grad=0, crop_ratio=0.875, data_dir='/home/ubuntu/yizhu/data/kinetics400/kinetics400/train_256', dataset='kinetics400', dtype='float32', eval=False, fast_temporal_stride=2, hard_weight=0.5, hashtag='', input_5d=False, input_size=224, kvstore=None, label_smoothing=False, last_gamma=False, log_interval=50, logging_file='i3d_inceptionv1_seg1_kinetics400_b8_g8_video_cleanv3.txt', lr=0.01, lr_decay=0.1, lr_decay_epoch='40,80,100', lr_decay_period=0, lr_mode='step', mixup=False, mixup_alpha=0.2, mixup_off_epoch=0, mode='hybrid', model='i3d_inceptionv1_kinetics400', momentum=0.9, new_height=256, new_length=32, new_step=2, new_width=340, no_wd=False, num_classes=400, num_crop=1, num_epochs=100, num_gpus=8, num_segments=1, num_workers=32, partial_bn=False, prefetch_ratio=1.0, resume_epoch=0, resume_params='', resume_states='', save_dir='/home/ubuntu/yizhu/logs/mxnet/kinetics400/i3d/i3d_inceptionv1_seg1_kinetics400_b8_g8_video_cleanv3', save_frequency=20, scale_ratios='1.0,0.8', slow_temporal_stride=16, slowfast=False, teacher=None, temperature=20, train_list='/home/ubuntu/yizhu/data/kinetics400/kinetics400/k400_train_240618.txt', use_amp=False, use_decord=True, use_gn=False, use_pretrained=False, use_se=False, use_tsn=False, val_data_dir='/home/ubuntu/yizhu/data/kinetics400/kinetics400/val_256', val_list='/home/ubuntu/yizhu/code/gluon-cv/extra/kinetics400/k400_val_19761_cleanv3.txt', video_loader=True, warmup_epochs=0, warmup_lr=0.0, wd=0.0001)
Total batch size is set to 64 on 8 GPUs
I3D_InceptionV1(
  (features): HybridSequential(
    (0): HybridSequential(
      (0): Conv3D(3 -> 64, kernel_size=(7, 7, 7), stride=(2, 2, 2), padding=(3, 3, 3), bias=False)
      (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=64)
      (2): Activation(relu)
    )
    (1): MaxPool3D(size=(1, 3, 3), stride=(1, 2, 2), padding=(0, 1, 1), ceil_mode=False, global_pool=False, pool_type=max, layout=NCDHW)
    (2): HybridSequential(
      (0): Conv3D(64 -> 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)
      (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=64)
      (2): Activation(relu)
    )
    (3): HybridSequential(
      (0): Conv3D(64 -> 192, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)
      (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=192)
      (2): Activation(relu)
    )
    (4): MaxPool3D(size=(1, 3, 3), stride=(1, 2, 2), padding=(0, 1, 1), ceil_mode=False, global_pool=False, pool_type=max, layout=NCDHW)
    (5): HybridConcurrent(
      (0): HybridSequential(
        (0): HybridSequential(
          (0): Conv3D(192 -> 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=64)
          (2): Activation(relu)
        )
      )
      (1): HybridSequential(
        (0): HybridSequential(
          (0): Conv3D(192 -> 96, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=96)
          (2): Activation(relu)
        )
        (1): HybridSequential(
          (0): Conv3D(96 -> 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=128)
          (2): Activation(relu)
        )
      )
      (2): HybridSequential(
        (0): HybridSequential(
          (0): Conv3D(192 -> 16, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=16)
          (2): Activation(relu)
        )
        (1): HybridSequential(
          (0): Conv3D(16 -> 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=32)
          (2): Activation(relu)
        )
      )
      (3): HybridSequential(
        (0): MaxPool3D(size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), ceil_mode=False, global_pool=False, pool_type=max, layout=NCDHW)
        (1): HybridSequential(
          (0): Conv3D(192 -> 32, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=32)
          (2): Activation(relu)
        )
      )
    )
    (6): HybridConcurrent(
      (0): HybridSequential(
        (0): HybridSequential(
          (0): Conv3D(256 -> 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=128)
          (2): Activation(relu)
        )
      )
      (1): HybridSequential(
        (0): HybridSequential(
          (0): Conv3D(256 -> 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=128)
          (2): Activation(relu)
        )
        (1): HybridSequential(
          (0): Conv3D(128 -> 192, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=192)
          (2): Activation(relu)
        )
      )
      (2): HybridSequential(
        (0): HybridSequential(
          (0): Conv3D(256 -> 32, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=32)
          (2): Activation(relu)
        )
        (1): HybridSequential(
          (0): Conv3D(32 -> 96, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=96)
          (2): Activation(relu)
        )
      )
      (3): HybridSequential(
        (0): MaxPool3D(size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), ceil_mode=False, global_pool=False, pool_type=max, layout=NCDHW)
        (1): HybridSequential(
          (0): Conv3D(256 -> 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=64)
          (2): Activation(relu)
        )
      )
    )
    (7): MaxPool3D(size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1), ceil_mode=False, global_pool=False, pool_type=max, layout=NCDHW)
    (8): HybridConcurrent(
      (0): HybridSequential(
        (0): HybridSequential(
          (0): Conv3D(480 -> 192, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=192)
          (2): Activation(relu)
        )
      )
      (1): HybridSequential(
        (0): HybridSequential(
          (0): Conv3D(480 -> 96, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=96)
          (2): Activation(relu)
        )
        (1): HybridSequential(
          (0): Conv3D(96 -> 208, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=208)
          (2): Activation(relu)
        )
      )
      (2): HybridSequential(
        (0): HybridSequential(
          (0): Conv3D(480 -> 16, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=16)
          (2): Activation(relu)
        )
        (1): HybridSequential(
          (0): Conv3D(16 -> 48, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=48)
          (2): Activation(relu)
        )
      )
      (3): HybridSequential(
        (0): MaxPool3D(size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), ceil_mode=False, global_pool=False, pool_type=max, layout=NCDHW)
        (1): HybridSequential(
          (0): Conv3D(480 -> 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=64)
          (2): Activation(relu)
        )
      )
    )
    (9): HybridConcurrent(
      (0): HybridSequential(
        (0): HybridSequential(
          (0): Conv3D(512 -> 160, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=160)
          (2): Activation(relu)
        )
      )
      (1): HybridSequential(
        (0): HybridSequential(
          (0): Conv3D(512 -> 112, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=112)
          (2): Activation(relu)
        )
        (1): HybridSequential(
          (0): Conv3D(112 -> 224, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=224)
          (2): Activation(relu)
        )
      )
      (2): HybridSequential(
        (0): HybridSequential(
          (0): Conv3D(512 -> 24, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=24)
          (2): Activation(relu)
        )
        (1): HybridSequential(
          (0): Conv3D(24 -> 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=64)
          (2): Activation(relu)
        )
      )
      (3): HybridSequential(
        (0): MaxPool3D(size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), ceil_mode=False, global_pool=False, pool_type=max, layout=NCDHW)
        (1): HybridSequential(
          (0): Conv3D(512 -> 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=64)
          (2): Activation(relu)
        )
      )
    )
    (10): HybridConcurrent(
      (0): HybridSequential(
        (0): HybridSequential(
          (0): Conv3D(512 -> 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=128)
          (2): Activation(relu)
        )
      )
      (1): HybridSequential(
        (0): HybridSequential(
          (0): Conv3D(512 -> 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=128)
          (2): Activation(relu)
        )
        (1): HybridSequential(
          (0): Conv3D(128 -> 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=256)
          (2): Activation(relu)
        )
      )
      (2): HybridSequential(
        (0): HybridSequential(
          (0): Conv3D(512 -> 24, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=24)
          (2): Activation(relu)
        )
        (1): HybridSequential(
          (0): Conv3D(24 -> 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=64)
          (2): Activation(relu)
        )
      )
      (3): HybridSequential(
        (0): MaxPool3D(size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), ceil_mode=False, global_pool=False, pool_type=max, layout=NCDHW)
        (1): HybridSequential(
          (0): Conv3D(512 -> 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=64)
          (2): Activation(relu)
        )
      )
    )
    (11): HybridConcurrent(
      (0): HybridSequential(
        (0): HybridSequential(
          (0): Conv3D(512 -> 112, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=112)
          (2): Activation(relu)
        )
      )
      (1): HybridSequential(
        (0): HybridSequential(
          (0): Conv3D(512 -> 144, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=144)
          (2): Activation(relu)
        )
        (1): HybridSequential(
          (0): Conv3D(144 -> 288, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=288)
          (2): Activation(relu)
        )
      )
      (2): HybridSequential(
        (0): HybridSequential(
          (0): Conv3D(512 -> 32, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=32)
          (2): Activation(relu)
        )
        (1): HybridSequential(
          (0): Conv3D(32 -> 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=64)
          (2): Activation(relu)
        )
      )
      (3): HybridSequential(
        (0): MaxPool3D(size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), ceil_mode=False, global_pool=False, pool_type=max, layout=NCDHW)
        (1): HybridSequential(
          (0): Conv3D(512 -> 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=64)
          (2): Activation(relu)
        )
      )
    )
    (12): HybridConcurrent(
      (0): HybridSequential(
        (0): HybridSequential(
          (0): Conv3D(528 -> 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=256)
          (2): Activation(relu)
        )
      )
      (1): HybridSequential(
        (0): HybridSequential(
          (0): Conv3D(528 -> 160, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=160)
          (2): Activation(relu)
        )
        (1): HybridSequential(
          (0): Conv3D(160 -> 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=320)
          (2): Activation(relu)
        )
      )
      (2): HybridSequential(
        (0): HybridSequential(
          (0): Conv3D(528 -> 32, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=32)
          (2): Activation(relu)
        )
        (1): HybridSequential(
          (0): Conv3D(32 -> 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=128)
          (2): Activation(relu)
        )
      )
      (3): HybridSequential(
        (0): MaxPool3D(size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), ceil_mode=False, global_pool=False, pool_type=max, layout=NCDHW)
        (1): HybridSequential(
          (0): Conv3D(528 -> 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=128)
          (2): Activation(relu)
        )
      )
    )
    (13): MaxPool3D(size=(2, 2, 2), stride=(2, 2, 2), padding=(0, 0, 0), ceil_mode=False, global_pool=False, pool_type=max, layout=NCDHW)
    (14): HybridConcurrent(
      (0): HybridSequential(
        (0): HybridSequential(
          (0): Conv3D(832 -> 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=256)
          (2): Activation(relu)
        )
      )
      (1): HybridSequential(
        (0): HybridSequential(
          (0): Conv3D(832 -> 160, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=160)
          (2): Activation(relu)
        )
        (1): HybridSequential(
          (0): Conv3D(160 -> 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=320)
          (2): Activation(relu)
        )
      )
      (2): HybridSequential(
        (0): HybridSequential(
          (0): Conv3D(832 -> 32, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=32)
          (2): Activation(relu)
        )
        (1): HybridSequential(
          (0): Conv3D(32 -> 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=128)
          (2): Activation(relu)
        )
      )
      (3): HybridSequential(
        (0): MaxPool3D(size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), ceil_mode=False, global_pool=False, pool_type=max, layout=NCDHW)
        (1): HybridSequential(
          (0): Conv3D(832 -> 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=128)
          (2): Activation(relu)
        )
      )
    )
    (15): HybridConcurrent(
      (0): HybridSequential(
        (0): HybridSequential(
          (0): Conv3D(832 -> 384, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=384)
          (2): Activation(relu)
        )
      )
      (1): HybridSequential(
        (0): HybridSequential(
          (0): Conv3D(832 -> 192, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=192)
          (2): Activation(relu)
        )
        (1): HybridSequential(
          (0): Conv3D(192 -> 384, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=384)
          (2): Activation(relu)
        )
      )
      (2): HybridSequential(
        (0): HybridSequential(
          (0): Conv3D(832 -> 48, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=48)
          (2): Activation(relu)
        )
        (1): HybridSequential(
          (0): Conv3D(48 -> 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=128)
          (2): Activation(relu)
        )
      )
      (3): HybridSequential(
        (0): MaxPool3D(size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), ceil_mode=False, global_pool=False, pool_type=max, layout=NCDHW)
        (1): HybridSequential(
          (0): Conv3D(832 -> 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)
          (1): BatchNorm(axis=1, eps=0.001, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=128)
          (2): Activation(relu)
        )
      )
    )
    (16): GlobalAvgPool3D(size=(1, 1, 1), stride=(1, 1, 1), padding=(0, 0, 0), ceil_mode=True, global_pool=True, pool_type=avg, layout=NCDHW)
  )
  (head): HybridSequential(
    (0): Dropout(p = 0.5, axes=())
    (1): Dense(1024 -> 400, linear)
  )
  (output): Dense(1024 -> 400, linear)
)
Load 240618 training samples and 19404 validation samples.
Epoch[000] Batch [0049]/[3759]	Speed: 36.840570 samples/sec	 accuracy=0.781250	 loss=5.964761	 lr=0.010000
Epoch[000] Batch [0099]/[3759]	Speed: 64.947627 samples/sec	 accuracy=1.062500	 loss=5.870239	 lr=0.010000
Epoch[000] Batch [0149]/[3759]	Speed: 71.111515 samples/sec	 accuracy=1.375000	 loss=5.782267	 lr=0.010000
Epoch[000] Batch [0199]/[3759]	Speed: 63.436096 samples/sec	 accuracy=1.789062	 loss=5.699661	 lr=0.010000
Epoch[000] Batch [0249]/[3759]	Speed: 69.565137 samples/sec	 accuracy=2.200000	 loss=5.624246	 lr=0.010000
Epoch[000] Batch [0299]/[3759]	Speed: 67.985561 samples/sec	 accuracy=2.588542	 loss=5.555615	 lr=0.010000
Epoch[000] Batch [0349]/[3759]	Speed: 63.217509 samples/sec	 accuracy=2.941964	 loss=5.491907	 lr=0.010000
Epoch[000] Batch [0399]/[3759]	Speed: 67.799570 samples/sec	 accuracy=3.289062	 loss=5.434089	 lr=0.010000
Epoch[000] Batch [0449]/[3759]	Speed: 68.640688 samples/sec	 accuracy=3.583333	 loss=5.389808	 lr=0.010000
Epoch[000] Batch [0499]/[3759]	Speed: 64.514312 samples/sec	 accuracy=3.856250	 loss=5.347190	 lr=0.010000
Epoch[000] Batch [0549]/[3759]	Speed: 67.443734 samples/sec	 accuracy=4.110795	 loss=5.305386	 lr=0.010000
Epoch[000] Batch [0599]/[3759]	Speed: 67.476828 samples/sec	 accuracy=4.393229	 loss=5.263992	 lr=0.010000
Epoch[000] Batch [0649]/[3759]	Speed: 67.436059 samples/sec	 accuracy=4.644231	 loss=5.225668	 lr=0.010000
Epoch[000] Batch [0699]/[3759]	Speed: 63.721610 samples/sec	 accuracy=4.906250	 loss=5.189578	 lr=0.010000
Epoch[000] Batch [0749]/[3759]	Speed: 65.168632 samples/sec	 accuracy=5.147917	 loss=5.154303	 lr=0.010000
Epoch[000] Batch [0799]/[3759]	Speed: 69.534883 samples/sec	 accuracy=5.373047	 loss=5.121004	 lr=0.010000
Epoch[000] Batch [0849]/[3759]	Speed: 69.327262 samples/sec	 accuracy=5.580882	 loss=5.090951	 lr=0.010000
Epoch[000] Batch [0899]/[3759]	Speed: 63.385556 samples/sec	 accuracy=5.840278	 loss=5.063207	 lr=0.010000
Epoch[000] Batch [0949]/[3759]	Speed: 69.835783 samples/sec	 accuracy=6.055921	 loss=5.035790	 lr=0.010000
Epoch[000] Batch [0999]/[3759]	Speed: 65.196955 samples/sec	 accuracy=6.246875	 loss=5.010366	 lr=0.010000
Epoch[000] Batch [1049]/[3759]	Speed: 68.939462 samples/sec	 accuracy=6.443452	 loss=4.985611	 lr=0.010000
Epoch[000] Batch [1099]/[3759]	Speed: 64.233892 samples/sec	 accuracy=6.697443	 loss=4.957331	 lr=0.010000
Epoch[000] Batch [1149]/[3759]	Speed: 68.409163 samples/sec	 accuracy=6.964674	 loss=4.932436	 lr=0.010000
Epoch[000] Batch [1199]/[3759]	Speed: 66.136422 samples/sec	 accuracy=7.184896	 loss=4.909750	 lr=0.010000
Epoch[000] Batch [1249]/[3759]	Speed: 66.500724 samples/sec	 accuracy=7.408750	 loss=4.887396	 lr=0.010000
Epoch[000] Batch [1299]/[3759]	Speed: 67.139294 samples/sec	 accuracy=7.599760	 loss=4.865990	 lr=0.010000
Epoch[000] Batch [1349]/[3759]	Speed: 68.503564 samples/sec	 accuracy=7.829861	 loss=4.843018	 lr=0.010000
Epoch[000] Batch [1399]/[3759]	Speed: 64.157302 samples/sec	 accuracy=8.037946	 loss=4.820966	 lr=0.010000
Epoch[000] Batch [1449]/[3759]	Speed: 67.787620 samples/sec	 accuracy=8.238147	 loss=4.800765	 lr=0.010000
Epoch[000] Batch [1499]/[3759]	Speed: 67.993547 samples/sec	 accuracy=8.426042	 loss=4.782592	 lr=0.010000
Epoch[000] Batch [1549]/[3759]	Speed: 65.524926 samples/sec	 accuracy=8.590726	 loss=4.764639	 lr=0.010000
Epoch[000] Batch [1599]/[3759]	Speed: 68.214348 samples/sec	 accuracy=8.773438	 loss=4.747445	 lr=0.010000
Epoch[000] Batch [1649]/[3759]	Speed: 64.104087 samples/sec	 accuracy=8.968750	 loss=4.730077	 lr=0.010000
Epoch[000] Batch [1699]/[3759]	Speed: 70.188200 samples/sec	 accuracy=9.136029	 loss=4.713666	 lr=0.010000
Epoch[000] Batch [1749]/[3759]	Speed: 67.470304 samples/sec	 accuracy=9.283036	 loss=4.698741	 lr=0.010000
Epoch[000] Batch [1799]/[3759]	Speed: 65.239379 samples/sec	 accuracy=9.447049	 loss=4.682890	 lr=0.010000
Epoch[000] Batch [1849]/[3759]	Speed: 64.869130 samples/sec	 accuracy=9.642736	 loss=4.667933	 lr=0.010000
Epoch[000] Batch [1899]/[3759]	Speed: 68.596384 samples/sec	 accuracy=9.784539	 loss=4.654010	 lr=0.010000
Epoch[000] Batch [1949]/[3759]	Speed: 66.372714 samples/sec	 accuracy=9.954327	 loss=4.639369	 lr=0.010000
Epoch[000] Batch [1999]/[3759]	Speed: 65.711512 samples/sec	 accuracy=10.089062	 loss=4.626295	 lr=0.010000
Epoch[000] Batch [2049]/[3759]	Speed: 68.174269 samples/sec	 accuracy=10.233232	 loss=4.612383	 lr=0.010000
Epoch[000] Batch [2099]/[3759]	Speed: 69.088729 samples/sec	 accuracy=10.402530	 loss=4.597914	 lr=0.010000
Epoch[000] Batch [2149]/[3759]	Speed: 64.618933 samples/sec	 accuracy=10.540698	 loss=4.585129	 lr=0.010000
Epoch[000] Batch [2199]/[3759]	Speed: 67.617235 samples/sec	 accuracy=10.691051	 loss=4.572411	 lr=0.010000
Epoch[000] Batch [2249]/[3759]	Speed: 67.444242 samples/sec	 accuracy=10.845139	 loss=4.559606	 lr=0.010000
Epoch[000] Batch [2299]/[3759]	Speed: 67.529655 samples/sec	 accuracy=11.022418	 loss=4.544884	 lr=0.010000
Epoch[000] Batch [2349]/[3759]	Speed: 64.435922 samples/sec	 accuracy=11.152261	 loss=4.532090	 lr=0.010000
Epoch[000] Batch [2399]/[3759]	Speed: 68.959364 samples/sec	 accuracy=11.305990	 loss=4.520189	 lr=0.010000
Epoch[000] Batch [2449]/[3759]	Speed: 68.598752 samples/sec	 accuracy=11.450893	 loss=4.507208	 lr=0.010000
Epoch[000] Batch [2499]/[3759]	Speed: 63.881305 samples/sec	 accuracy=11.589375	 loss=4.494840	 lr=0.010000
Epoch[000] Batch [2549]/[3759]	Speed: 68.325107 samples/sec	 accuracy=11.741422	 loss=4.483142	 lr=0.010000
Epoch[000] Batch [2599]/[3759]	Speed: 64.256684 samples/sec	 accuracy=11.880409	 loss=4.472178	 lr=0.010000
Epoch[000] Batch [2649]/[3759]	Speed: 69.570584 samples/sec	 accuracy=12.010024	 loss=4.460262	 lr=0.010000
Epoch[000] Batch [2699]/[3759]	Speed: 66.580432 samples/sec	 accuracy=12.135417	 loss=4.449342	 lr=0.010000
Epoch[000] Batch [2749]/[3759]	Speed: 66.831497 samples/sec	 accuracy=12.273864	 loss=4.438059	 lr=0.010000
Epoch[000] Batch [2799]/[3759]	Speed: 64.868470 samples/sec	 accuracy=12.402902	 loss=4.427469	 lr=0.010000
Epoch[000] Batch [2849]/[3759]	Speed: 68.660284 samples/sec	 accuracy=12.540570	 loss=4.416441	 lr=0.010000
Epoch[000] Batch [2899]/[3759]	Speed: 65.327005 samples/sec	 accuracy=12.666487	 loss=4.406608	 lr=0.010000
Epoch[000] Batch [2949]/[3759]	Speed: 66.943213 samples/sec	 accuracy=12.784958	 loss=4.395522	 lr=0.010000
Epoch[000] Batch [2999]/[3759]	Speed: 67.582216 samples/sec	 accuracy=12.913021	 loss=4.385482	 lr=0.010000
Epoch[000] Batch [3049]/[3759]	Speed: 65.774736 samples/sec	 accuracy=13.040984	 loss=4.376118	 lr=0.010000
Epoch[000] Batch [3099]/[3759]	Speed: 65.614130 samples/sec	 accuracy=13.164315	 loss=4.366091	 lr=0.010000
Epoch[000] Batch [3149]/[3759]	Speed: 68.929254 samples/sec	 accuracy=13.292163	 loss=4.356373	 lr=0.010000
Epoch[000] Batch [3199]/[3759]	Speed: 67.077510 samples/sec	 accuracy=13.420410	 loss=4.346997	 lr=0.010000
Epoch[000] Batch [3249]/[3759]	Speed: 64.475402 samples/sec	 accuracy=13.523077	 loss=4.338123	 lr=0.010000
Epoch[000] Batch [3299]/[3759]	Speed: 68.197253 samples/sec	 accuracy=13.640625	 loss=4.328605	 lr=0.010000
Epoch[000] Batch [3349]/[3759]	Speed: 66.089170 samples/sec	 accuracy=13.748134	 loss=4.320654	 lr=0.010000
Epoch[000] Batch [3399]/[3759]	Speed: 67.940605 samples/sec	 accuracy=13.837316	 loss=4.312175	 lr=0.010000
Epoch[000] Batch [3449]/[3759]	Speed: 68.136956 samples/sec	 accuracy=13.941576	 loss=4.303860	 lr=0.010000
Epoch[000] Batch [3499]/[3759]	Speed: 65.655432 samples/sec	 accuracy=14.051339	 loss=4.294730	 lr=0.010000
Epoch[000] Batch [3549]/[3759]	Speed: 66.589288 samples/sec	 accuracy=14.147887	 loss=4.287042	 lr=0.010000
Epoch[000] Batch [3599]/[3759]	Speed: 63.588087 samples/sec	 accuracy=14.250868	 loss=4.278162	 lr=0.010000
Epoch[000] Batch [3649]/[3759]	Speed: 68.807954 samples/sec	 accuracy=14.347603	 loss=4.270259	 lr=0.010000
Epoch[000] Batch [3699]/[3759]	Speed: 68.185288 samples/sec	 accuracy=14.459037	 loss=4.261687	 lr=0.010000
Epoch[000] Batch [3749]/[3759]	Speed: 74.586626 samples/sec	 accuracy=14.550833	 loss=4.254391	 lr=0.010000
Batch [0049]/[0303]: acc-top1=23.750000 acc-top5=48.187500
Batch [0099]/[0303]: acc-top1=23.687500 acc-top5=48.687500
Batch [0149]/[0303]: acc-top1=23.833333 acc-top5=49.135417
Batch [0199]/[0303]: acc-top1=24.085938 acc-top5=49.570312
Batch [0249]/[0303]: acc-top1=24.087500 acc-top5=49.693750
Batch [0299]/[0303]: acc-top1=23.802083 acc-top5=49.536458
[Epoch 000] training: accuracy=14.565875	 loss=4.253158
[Epoch 000] speed: 66 samples/sec	time cost: 3944.831826
[Epoch 000] validation: acc-top1=23.819101 acc-top5=49.525578 loss=3.683455
Epoch[001] Batch [0049]/[3760]	Speed: 45.551238 samples/sec	 accuracy=22.187500	 loss=3.628117	 lr=0.010000
Epoch[001] Batch [0099]/[3760]	Speed: 68.302459 samples/sec	 accuracy=22.453125	 loss=3.607098	 lr=0.010000
Epoch[001] Batch [0149]/[3760]	Speed: 66.875441 samples/sec	 accuracy=22.625000	 loss=3.605632	 lr=0.010000
Epoch[001] Batch [0199]/[3760]	Speed: 67.557002 samples/sec	 accuracy=22.710938	 loss=3.600579	 lr=0.010000
Epoch[001] Batch [0249]/[3760]	Speed: 67.185587 samples/sec	 accuracy=22.825000	 loss=3.590850	 lr=0.010000
Epoch[001] Batch [0299]/[3760]	Speed: 66.068549 samples/sec	 accuracy=22.822917	 loss=3.590546	 lr=0.010000
Epoch[001] Batch [0349]/[3760]	Speed: 67.992725 samples/sec	 accuracy=23.071429	 loss=3.581594	 lr=0.010000
Epoch[001] Batch [0399]/[3760]	Speed: 63.452975 samples/sec	 accuracy=23.031250	 loss=3.577785	 lr=0.010000
Epoch[001] Batch [0449]/[3760]	Speed: 69.476013 samples/sec	 accuracy=23.052083	 loss=3.578720	 lr=0.010000
Epoch[001] Batch [0499]/[3760]	Speed: 66.746772 samples/sec	 accuracy=23.021875	 loss=3.580720	 lr=0.010000
Epoch[001] Batch [0549]/[3760]	Speed: 66.440365 samples/sec	 accuracy=22.960227	 loss=3.578635	 lr=0.010000
Epoch[001] Batch [0599]/[3760]	Speed: 66.259872 samples/sec	 accuracy=23.075521	 loss=3.571297	 lr=0.010000
Epoch[001] Batch [0649]/[3760]	Speed: 65.856452 samples/sec	 accuracy=23.088942	 loss=3.572548	 lr=0.010000
Epoch[001] Batch [0699]/[3760]	Speed: 66.413989 samples/sec	 accuracy=23.158482	 loss=3.570543	 lr=0.010000
Epoch[001] Batch [0749]/[3760]	Speed: 66.017473 samples/sec	 accuracy=23.258333	 loss=3.567609	 lr=0.010000
Epoch[001] Batch [0799]/[3760]	Speed: 68.596365 samples/sec	 accuracy=23.292969	 loss=3.564966	 lr=0.010000
Epoch[001] Batch [0849]/[3760]	Speed: 68.700003 samples/sec	 accuracy=23.329044	 loss=3.562320	 lr=0.010000
Epoch[001] Batch [0899]/[3760]	Speed: 64.381095 samples/sec	 accuracy=23.395833	 loss=3.558422	 lr=0.010000
Epoch[001] Batch [0949]/[3760]	Speed: 69.512213 samples/sec	 accuracy=23.509868	 loss=3.552302	 lr=0.010000
Epoch[001] Batch [0999]/[3760]	Speed: 65.621930 samples/sec	 accuracy=23.540625	 loss=3.548851	 lr=0.010000
Epoch[001] Batch [1049]/[3760]	Speed: 65.728487 samples/sec	 accuracy=23.572917	 loss=3.549019	 lr=0.010000
Epoch[001] Batch [1099]/[3760]	Speed: 68.237105 samples/sec	 accuracy=23.595170	 loss=3.549583	 lr=0.010000
Epoch[001] Batch [1149]/[3760]	Speed: 67.615230 samples/sec	 accuracy=23.650815	 loss=3.547044	 lr=0.010000
Epoch[001] Batch [1199]/[3760]	Speed: 66.131884 samples/sec	 accuracy=23.697917	 loss=3.544270	 lr=0.010000
Epoch[001] Batch [1249]/[3760]	Speed: 67.433856 samples/sec	 accuracy=23.735000	 loss=3.541105	 lr=0.010000
Epoch[001] Batch [1299]/[3760]	Speed: 65.230842 samples/sec	 accuracy=23.781250	 loss=3.540533	 lr=0.010000
Epoch[001] Batch [1349]/[3760]	Speed: 65.759182 samples/sec	 accuracy=23.849537	 loss=3.537565	 lr=0.010000
Epoch[001] Batch [1399]/[3760]	Speed: 67.457366 samples/sec	 accuracy=23.872768	 loss=3.536060	 lr=0.010000
Epoch[001] Batch [1449]/[3760]	Speed: 66.429224 samples/sec	 accuracy=23.891164	 loss=3.533373	 lr=0.010000
Epoch[001] Batch [1499]/[3760]	Speed: 66.740326 samples/sec	 accuracy=23.958333	 loss=3.528870	 lr=0.010000
Epoch[001] Batch [1549]/[3760]	Speed: 67.995755 samples/sec	 accuracy=23.997984	 loss=3.525869	 lr=0.010000
Epoch[001] Batch [1599]/[3760]	Speed: 64.899452 samples/sec	 accuracy=24.060547	 loss=3.521891	 lr=0.010000
Epoch[001] Batch [1649]/[3760]	Speed: 69.288956 samples/sec	 accuracy=24.100379	 loss=3.520385	 lr=0.010000
Epoch[001] Batch [1699]/[3760]	Speed: 68.512924 samples/sec	 accuracy=24.115809	 loss=3.518894	 lr=0.010000
Epoch[001] Batch [1749]/[3760]	Speed: 64.248300 samples/sec	 accuracy=24.191071	 loss=3.515379	 lr=0.010000
Epoch[001] Batch [1799]/[3760]	Speed: 67.723726 samples/sec	 accuracy=24.234375	 loss=3.512511	 lr=0.010000
Epoch[001] Batch [1849]/[3760]	Speed: 65.666440 samples/sec	 accuracy=24.263514	 loss=3.509976	 lr=0.010000
Epoch[001] Batch [1899]/[3760]	Speed: 68.129514 samples/sec	 accuracy=24.309211	 loss=3.507195	 lr=0.010000
Epoch[001] Batch [1949]/[3760]	Speed: 66.683728 samples/sec	 accuracy=24.378205	 loss=3.503892	 lr=0.010000
Epoch[001] Batch [1999]/[3760]	Speed: 67.659607 samples/sec	 accuracy=24.412500	 loss=3.502624	 lr=0.010000
Epoch[001] Batch [2049]/[3760]	Speed: 64.428645 samples/sec	 accuracy=24.480183	 loss=3.498343	 lr=0.010000
Epoch[001] Batch [2099]/[3760]	Speed: 68.781046 samples/sec	 accuracy=24.553571	 loss=3.494656	 lr=0.010000
Epoch[001] Batch [2149]/[3760]	Speed: 66.650918 samples/sec	 accuracy=24.595930	 loss=3.492170	 lr=0.010000
Epoch[001] Batch [2199]/[3760]	Speed: 66.031571 samples/sec	 accuracy=24.639205	 loss=3.489470	 lr=0.010000
Epoch[001] Batch [2249]/[3760]	Speed: 69.069732 samples/sec	 accuracy=24.676389	 loss=3.487030	 lr=0.010000
Epoch[001] Batch [2299]/[3760]	Speed: 66.672092 samples/sec	 accuracy=24.729620	 loss=3.484413	 lr=0.010000
Epoch[001] Batch [2349]/[3760]	Speed: 65.693627 samples/sec	 accuracy=24.776596	 loss=3.481568	 lr=0.010000
Epoch[001] Batch [2399]/[3760]	Speed: 68.098197 samples/sec	 accuracy=24.824219	 loss=3.478715	 lr=0.010000
Epoch[001] Batch [2449]/[3760]	Speed: 68.111854 samples/sec	 accuracy=24.876276	 loss=3.475459	 lr=0.010000
Epoch[001] Batch [2499]/[3760]	Speed: 64.581125 samples/sec	 accuracy=24.933125	 loss=3.472347	 lr=0.010000
Epoch[001] Batch [2549]/[3760]	Speed: 67.039346 samples/sec	 accuracy=24.984681	 loss=3.468573	 lr=0.010000
Epoch[001] Batch [2599]/[3760]	Speed: 65.504244 samples/sec	 accuracy=25.010216	 loss=3.466111	 lr=0.010000
Epoch[001] Batch [2649]/[3760]	Speed: 67.795537 samples/sec	 accuracy=25.053066	 loss=3.464094	 lr=0.010000
Epoch[001] Batch [2699]/[3760]	Speed: 64.915774 samples/sec	 accuracy=25.103009	 loss=3.460570	 lr=0.010000
Epoch[001] Batch [2749]/[3760]	Speed: 68.544639 samples/sec	 accuracy=25.138068	 loss=3.457989	 lr=0.010000
Epoch[001] Batch [2799]/[3760]	Speed: 67.826404 samples/sec	 accuracy=25.207031	 loss=3.454580	 lr=0.010000
Epoch[001] Batch [2849]/[3760]	Speed: 65.266867 samples/sec	 accuracy=25.273575	 loss=3.451178	 lr=0.010000
Epoch[001] Batch [2899]/[3760]	Speed: 67.759944 samples/sec	 accuracy=25.329741	 loss=3.448059	 lr=0.010000
Epoch[001] Batch [2949]/[3760]	Speed: 64.695047 samples/sec	 accuracy=25.399894	 loss=3.444400	 lr=0.010000
Epoch[001] Batch [2999]/[3760]	Speed: 69.874046 samples/sec	 accuracy=25.446354	 loss=3.441898	 lr=0.010000
Epoch[001] Batch [3049]/[3760]	Speed: 68.332432 samples/sec	 accuracy=25.501025	 loss=3.438723	 lr=0.010000
Epoch[001] Batch [3099]/[3760]	Speed: 65.202264 samples/sec	 accuracy=25.537298	 loss=3.436072	 lr=0.010000
Epoch[001] Batch [3149]/[3760]	Speed: 66.096256 samples/sec	 accuracy=25.600694	 loss=3.432206	 lr=0.010000
Epoch[001] Batch [3199]/[3760]	Speed: 67.277714 samples/sec	 accuracy=25.625000	 loss=3.430490	 lr=0.010000
Epoch[001] Batch [3249]/[3760]	Speed: 66.815005 samples/sec	 accuracy=25.673558	 loss=3.427256	 lr=0.010000
Epoch[001] Batch [3299]/[3760]	Speed: 68.984295 samples/sec	 accuracy=25.697443	 loss=3.424711	 lr=0.010000
Epoch[001] Batch [3349]/[3760]	Speed: 65.147968 samples/sec	 accuracy=25.736474	 loss=3.421761	 lr=0.010000
Epoch[001] Batch [3399]/[3760]	Speed: 68.221854 samples/sec	 accuracy=25.762408	 loss=3.419839	 lr=0.010000
Epoch[001] Batch [3449]/[3760]	Speed: 64.732124 samples/sec	 accuracy=25.817029	 loss=3.417087	 lr=0.010000
Epoch[001] Batch [3499]/[3760]	Speed: 67.822100 samples/sec	 accuracy=25.841518	 loss=3.415339	 lr=0.010000
Epoch[001] Batch [3549]/[3760]	Speed: 65.437702 samples/sec	 accuracy=25.861356	 loss=3.413728	 lr=0.010000
Epoch[001] Batch [3599]/[3760]	Speed: 68.908776 samples/sec	 accuracy=25.900608	 loss=3.412030	 lr=0.010000
Epoch[001] Batch [3649]/[3760]	Speed: 68.469592 samples/sec	 accuracy=25.955051	 loss=3.409155	 lr=0.010000
Epoch[001] Batch [3699]/[3760]	Speed: 64.138950 samples/sec	 accuracy=25.995355	 loss=3.407165	 lr=0.010000
Epoch[001] Batch [3749]/[3760]	Speed: 74.256291 samples/sec	 accuracy=26.036667	 loss=3.404839	 lr=0.010000
Batch [0049]/[0303]: acc-top1=31.781250 acc-top5=59.781250
Batch [0099]/[0303]: acc-top1=31.437500 acc-top5=59.281250
Batch [0149]/[0303]: acc-top1=31.833333 acc-top5=59.875000
Batch [0199]/[0303]: acc-top1=32.335938 acc-top5=60.046875
Batch [0249]/[0303]: acc-top1=31.912500 acc-top5=60.212500
Batch [0299]/[0303]: acc-top1=31.703125 acc-top5=59.786458
[Epoch 001] training: accuracy=26.041805	 loss=3.404929
[Epoch 001] speed: 66 samples/sec	time cost: 3924.091269
[Epoch 001] validation: acc-top1=31.719266 acc-top5=59.777228 loss=3.176292
Epoch[002] Batch [0049]/[3759]	Speed: 45.173334 samples/sec	 accuracy=29.093750	 loss=3.209565	 lr=0.010000
Epoch[002] Batch [0099]/[3759]	Speed: 68.152135 samples/sec	 accuracy=30.046875	 loss=3.185053	 lr=0.010000
Epoch[002] Batch [0149]/[3759]	Speed: 70.123082 samples/sec	 accuracy=29.479167	 loss=3.179691	 lr=0.010000
Epoch[002] Batch [0199]/[3759]	Speed: 63.510858 samples/sec	 accuracy=29.773438	 loss=3.169317	 lr=0.010000
Epoch[002] Batch [0249]/[3759]	Speed: 68.914978 samples/sec	 accuracy=29.756250	 loss=3.167561	 lr=0.010000
Epoch[002] Batch [0299]/[3759]	Speed: 67.396787 samples/sec	 accuracy=29.750000	 loss=3.171641	 lr=0.010000
Epoch[002] Batch [0349]/[3759]	Speed: 65.785999 samples/sec	 accuracy=29.955357	 loss=3.166917	 lr=0.010000
Epoch[002] Batch [0399]/[3759]	Speed: 68.716303 samples/sec	 accuracy=29.906250	 loss=3.169808	 lr=0.010000
Epoch[002] Batch [0449]/[3759]	Speed: 67.469397 samples/sec	 accuracy=29.899306	 loss=3.164951	 lr=0.010000
Epoch[002] Batch [0499]/[3759]	Speed: 64.899401 samples/sec	 accuracy=30.018750	 loss=3.163133	 lr=0.010000
Epoch[002] Batch [0549]/[3759]	Speed: 68.312113 samples/sec	 accuracy=30.102273	 loss=3.160691	 lr=0.010000
Epoch[002] Batch [0599]/[3759]	Speed: 67.226959 samples/sec	 accuracy=30.036458	 loss=3.162648	 lr=0.010000
Epoch[002] Batch [0649]/[3759]	Speed: 65.124192 samples/sec	 accuracy=30.004808	 loss=3.166305	 lr=0.010000
Epoch[002] Batch [0699]/[3759]	Speed: 69.845787 samples/sec	 accuracy=30.062500	 loss=3.163379	 lr=0.010000
Epoch[002] Batch [0749]/[3759]	Speed: 64.243741 samples/sec	 accuracy=30.039583	 loss=3.163403	 lr=0.010000
Epoch[002] Batch [0799]/[3759]	Speed: 67.225328 samples/sec	 accuracy=30.101562	 loss=3.162859	 lr=0.010000
Epoch[002] Batch [0849]/[3759]	Speed: 66.844936 samples/sec	 accuracy=30.053309	 loss=3.160414	 lr=0.010000
Epoch[002] Batch [0899]/[3759]	Speed: 67.272045 samples/sec	 accuracy=30.065972	 loss=3.159837	 lr=0.010000
Epoch[002] Batch [0949]/[3759]	Speed: 65.946116 samples/sec	 accuracy=30.110197	 loss=3.156727	 lr=0.010000
Epoch[002] Batch [0999]/[3759]	Speed: 67.197319 samples/sec	 accuracy=30.253125	 loss=3.151797	 lr=0.010000
Epoch[002] Batch [1049]/[3759]	Speed: 67.560594 samples/sec	 accuracy=30.386905	 loss=3.144728	 lr=0.010000
Epoch[002] Batch [1099]/[3759]	Speed: 67.359955 samples/sec	 accuracy=30.362216	 loss=3.143836	 lr=0.010000
Epoch[002] Batch [1149]/[3759]	Speed: 64.887555 samples/sec	 accuracy=30.415761	 loss=3.142067	 lr=0.010000
Epoch[002] Batch [1199]/[3759]	Speed: 65.932377 samples/sec	 accuracy=30.506510	 loss=3.140512	 lr=0.010000
Epoch[002] Batch [1249]/[3759]	Speed: 66.262849 samples/sec	 accuracy=30.497500	 loss=3.141280	 lr=0.010000
Epoch[002] Batch [1299]/[3759]	Speed: 68.772015 samples/sec	 accuracy=30.518029	 loss=3.140962	 lr=0.010000
Epoch[002] Batch [1349]/[3759]	Speed: 68.197639 samples/sec	 accuracy=30.519676	 loss=3.140122	 lr=0.010000
Epoch[002] Batch [1399]/[3759]	Speed: 64.346104 samples/sec	 accuracy=30.495536	 loss=3.142523	 lr=0.010000
Epoch[002] Batch [1449]/[3759]	Speed: 67.858694 samples/sec	 accuracy=30.522629	 loss=3.142136	 lr=0.010000
Epoch[002] Batch [1499]/[3759]	Speed: 67.717430 samples/sec	 accuracy=30.625000	 loss=3.138527	 lr=0.010000
Epoch[002] Batch [1549]/[3759]	Speed: 67.071152 samples/sec	 accuracy=30.666331	 loss=3.138070	 lr=0.010000
Epoch[002] Batch [1599]/[3759]	Speed: 65.336756 samples/sec	 accuracy=30.685547	 loss=3.137228	 lr=0.010000
Epoch[002] Batch [1649]/[3759]	Speed: 67.559557 samples/sec	 accuracy=30.662879	 loss=3.136874	 lr=0.010000
Epoch[002] Batch [1699]/[3759]	Speed: 67.252106 samples/sec	 accuracy=30.670037	 loss=3.135543	 lr=0.010000
Epoch[002] Batch [1749]/[3759]	Speed: 63.480202 samples/sec	 accuracy=30.701786	 loss=3.134306	 lr=0.010000
Epoch[002] Batch [1799]/[3759]	Speed: 68.975428 samples/sec	 accuracy=30.719618	 loss=3.133234	 lr=0.010000
Epoch[002] Batch [1849]/[3759]	Speed: 67.155228 samples/sec	 accuracy=30.779561	 loss=3.130019	 lr=0.010000
Epoch[002] Batch [1899]/[3759]	Speed: 65.737782 samples/sec	 accuracy=30.807566	 loss=3.127740	 lr=0.010000
Epoch[002] Batch [1949]/[3759]	Speed: 69.314796 samples/sec	 accuracy=30.826122	 loss=3.125914	 lr=0.010000
Epoch[002] Batch [1999]/[3759]	Speed: 65.278859 samples/sec	 accuracy=30.827344	 loss=3.125436	 lr=0.010000
Epoch[002] Batch [2049]/[3759]	Speed: 68.154821 samples/sec	 accuracy=30.821646	 loss=3.124099	 lr=0.010000
Epoch[002] Batch [2099]/[3759]	Speed: 67.115892 samples/sec	 accuracy=30.848214	 loss=3.123104	 lr=0.010000
Epoch[002] Batch [2149]/[3759]	Speed: 66.279125 samples/sec	 accuracy=30.898983	 loss=3.120161	 lr=0.010000
Epoch[002] Batch [2199]/[3759]	Speed: 67.564330 samples/sec	 accuracy=30.916193	 loss=3.117942	 lr=0.010000
Epoch[002] Batch [2249]/[3759]	Speed: 67.988162 samples/sec	 accuracy=30.998611	 loss=3.114367	 lr=0.010000
Epoch[002] Batch [2299]/[3759]	Speed: 64.876706 samples/sec	 accuracy=31.036685	 loss=3.112280	 lr=0.010000
Epoch[002] Batch [2349]/[3759]	Speed: 69.143579 samples/sec	 accuracy=31.088431	 loss=3.109970	 lr=0.010000
Epoch[002] Batch [2399]/[3759]	Speed: 67.896343 samples/sec	 accuracy=31.114583	 loss=3.108016	 lr=0.010000
Epoch[002] Batch [2449]/[3759]	Speed: 64.651906 samples/sec	 accuracy=31.158163	 loss=3.106530	 lr=0.010000
Epoch[002] Batch [2499]/[3759]	Speed: 69.320922 samples/sec	 accuracy=31.208750	 loss=3.103952	 lr=0.010000
Epoch[002] Batch [2549]/[3759]	Speed: 65.662849 samples/sec	 accuracy=31.209559	 loss=3.102731	 lr=0.010000
Epoch[002] Batch [2599]/[3759]	Speed: 67.013765 samples/sec	 accuracy=31.230769	 loss=3.101983	 lr=0.010000
Epoch[002] Batch [2649]/[3759]	Speed: 68.012113 samples/sec	 accuracy=31.250590	 loss=3.100183	 lr=0.010000
Epoch[002] Batch [2699]/[3759]	Speed: 66.199961 samples/sec	 accuracy=31.268519	 loss=3.098257	 lr=0.010000
Epoch[002] Batch [2749]/[3759]	Speed: 67.023834 samples/sec	 accuracy=31.305114	 loss=3.096436	 lr=0.010000
Epoch[002] Batch [2799]/[3759]	Speed: 63.335582 samples/sec	 accuracy=31.323661	 loss=3.095852	 lr=0.010000
Epoch[002] Batch [2849]/[3759]	Speed: 69.656297 samples/sec	 accuracy=31.335526	 loss=3.095602	 lr=0.010000
Epoch[002] Batch [2899]/[3759]	Speed: 67.533892 samples/sec	 accuracy=31.342672	 loss=3.094844	 lr=0.010000
Epoch[002] Batch [2949]/[3759]	Speed: 63.878114 samples/sec	 accuracy=31.372352	 loss=3.092843	 lr=0.010000
Epoch[002] Batch [2999]/[3759]	Speed: 69.052916 samples/sec	 accuracy=31.415625	 loss=3.091625	 lr=0.010000
Epoch[002] Batch [3049]/[3759]	Speed: 67.929270 samples/sec	 accuracy=31.451844	 loss=3.089734	 lr=0.010000
Epoch[002] Batch [3099]/[3759]	Speed: 66.224960 samples/sec	 accuracy=31.441028	 loss=3.089952	 lr=0.010000
Epoch[002] Batch [3149]/[3759]	Speed: 66.100781 samples/sec	 accuracy=31.467262	 loss=3.088647	 lr=0.010000
Epoch[002] Batch [3199]/[3759]	Speed: 65.469422 samples/sec	 accuracy=31.511230	 loss=3.086858	 lr=0.010000
Epoch[002] Batch [3249]/[3759]	Speed: 70.069866 samples/sec	 accuracy=31.533173	 loss=3.085973	 lr=0.010000
Epoch[002] Batch [3299]/[3759]	Speed: 63.939179 samples/sec	 accuracy=31.543561	 loss=3.084338	 lr=0.010000
Epoch[002] Batch [3349]/[3759]	Speed: 68.906849 samples/sec	 accuracy=31.554571	 loss=3.083614	 lr=0.010000
Epoch[002] Batch [3399]/[3759]	Speed: 65.575412 samples/sec	 accuracy=31.601563	 loss=3.081506	 lr=0.010000
Epoch[002] Batch [3449]/[3759]	Speed: 67.078437 samples/sec	 accuracy=31.615942	 loss=3.080039	 lr=0.010000
Epoch[002] Batch [3499]/[3759]	Speed: 68.766673 samples/sec	 accuracy=31.638393	 loss=3.078747	 lr=0.010000
Epoch[002] Batch [3549]/[3759]	Speed: 63.154738 samples/sec	 accuracy=31.665933	 loss=3.077170	 lr=0.010000
Epoch[002] Batch [3599]/[3759]	Speed: 69.439218 samples/sec	 accuracy=31.694010	 loss=3.075032	 lr=0.010000
Epoch[002] Batch [3649]/[3759]	Speed: 65.412691 samples/sec	 accuracy=31.720462	 loss=3.074648	 lr=0.010000
Epoch[002] Batch [3699]/[3759]	Speed: 66.338905 samples/sec	 accuracy=31.753378	 loss=3.072937	 lr=0.010000
Epoch[002] Batch [3749]/[3759]	Speed: 75.154116 samples/sec	 accuracy=31.772500	 loss=3.071560	 lr=0.010000
Batch [0049]/[0303]: acc-top1=36.625000 acc-top5=65.000000
Batch [0099]/[0303]: acc-top1=36.171875 acc-top5=65.250000
Batch [0149]/[0303]: acc-top1=36.739583 acc-top5=65.343750
Batch [0199]/[0303]: acc-top1=37.187500 acc-top5=65.648438
Batch [0249]/[0303]: acc-top1=36.806250 acc-top5=65.562500
Batch [0299]/[0303]: acc-top1=36.645833 acc-top5=65.177083
[Epoch 002] training: accuracy=31.775821	 loss=3.071343
[Epoch 002] speed: 66 samples/sec	time cost: 3924.271649
[Epoch 002] validation: acc-top1=36.618193 acc-top5=65.114480 loss=2.868469
Epoch[003] Batch [0049]/[3760]	Speed: 46.863842 samples/sec	 accuracy=34.500000	 loss=2.926053	 lr=0.010000
Epoch[003] Batch [0099]/[3760]	Speed: 63.908579 samples/sec	 accuracy=34.140625	 loss=2.930776	 lr=0.010000
Epoch[003] Batch [0149]/[3760]	Speed: 68.486021 samples/sec	 accuracy=34.458333	 loss=2.937043	 lr=0.010000
Epoch[003] Batch [0199]/[3760]	Speed: 66.955871 samples/sec	 accuracy=34.625000	 loss=2.931509	 lr=0.010000
Epoch[003] Batch [0249]/[3760]	Speed: 64.422109 samples/sec	 accuracy=34.387500	 loss=2.933740	 lr=0.010000
Epoch[003] Batch [0299]/[3760]	Speed: 67.047083 samples/sec	 accuracy=34.531250	 loss=2.924623	 lr=0.010000
Epoch[003] Batch [0349]/[3760]	Speed: 68.594250 samples/sec	 accuracy=34.526786	 loss=2.924198	 lr=0.010000
Epoch[003] Batch [0399]/[3760]	Speed: 66.659849 samples/sec	 accuracy=34.421875	 loss=2.920473	 lr=0.010000
Epoch[003] Batch [0449]/[3760]	Speed: 66.745354 samples/sec	 accuracy=34.579861	 loss=2.911097	 lr=0.010000
Epoch[003] Batch [0499]/[3760]	Speed: 66.561626 samples/sec	 accuracy=34.659375	 loss=2.910402	 lr=0.010000
Epoch[003] Batch [0549]/[3760]	Speed: 68.601616 samples/sec	 accuracy=34.821023	 loss=2.903083	 lr=0.010000
Epoch[003] Batch [0599]/[3760]	Speed: 67.197284 samples/sec	 accuracy=34.864583	 loss=2.903082	 lr=0.010000
Epoch[003] Batch [0649]/[3760]	Speed: 63.479043 samples/sec	 accuracy=34.896635	 loss=2.901273	 lr=0.010000
Epoch[003] Batch [0699]/[3760]	Speed: 68.060717 samples/sec	 accuracy=34.843750	 loss=2.902948	 lr=0.010000
Epoch[003] Batch [0749]/[3760]	Speed: 65.775263 samples/sec	 accuracy=34.822917	 loss=2.903115	 lr=0.010000
Epoch[003] Batch [0799]/[3760]	Speed: 65.639689 samples/sec	 accuracy=34.826172	 loss=2.902924	 lr=0.010000
Epoch[003] Batch [0849]/[3760]	Speed: 68.702425 samples/sec	 accuracy=34.812500	 loss=2.901483	 lr=0.010000
Epoch[003] Batch [0899]/[3760]	Speed: 65.465832 samples/sec	 accuracy=34.855903	 loss=2.898987	 lr=0.010000
Epoch[003] Batch [0949]/[3760]	Speed: 69.622946 samples/sec	 accuracy=34.797697	 loss=2.903161	 lr=0.010000
Epoch[003] Batch [0999]/[3760]	Speed: 63.008887 samples/sec	 accuracy=34.801563	 loss=2.902091	 lr=0.010000
Epoch[003] Batch [1049]/[3760]	Speed: 68.760244 samples/sec	 accuracy=34.794643	 loss=2.902308	 lr=0.010000
Epoch[003] Batch [1099]/[3760]	Speed: 66.310191 samples/sec	 accuracy=34.853693	 loss=2.899495	 lr=0.010000
Epoch[003] Batch [1149]/[3760]	Speed: 65.834483 samples/sec	 accuracy=34.804348	 loss=2.902425	 lr=0.010000
Epoch[003] Batch [1199]/[3760]	Speed: 68.192336 samples/sec	 accuracy=34.799479	 loss=2.901219	 lr=0.010000
Epoch[003] Batch [1249]/[3760]	Speed: 67.575242 samples/sec	 accuracy=34.776250	 loss=2.901694	 lr=0.010000
Epoch[003] Batch [1299]/[3760]	Speed: 64.282914 samples/sec	 accuracy=34.817308	 loss=2.899387	 lr=0.010000
Epoch[003] Batch [1349]/[3760]	Speed: 68.180283 samples/sec	 accuracy=34.798611	 loss=2.898446	 lr=0.010000
Epoch[003] Batch [1399]/[3760]	Speed: 67.378478 samples/sec	 accuracy=34.780134	 loss=2.898791	 lr=0.010000
Epoch[003] Batch [1449]/[3760]	Speed: 65.720268 samples/sec	 accuracy=34.842672	 loss=2.895832	 lr=0.010000
Epoch[003] Batch [1499]/[3760]	Speed: 66.643746 samples/sec	 accuracy=34.892708	 loss=2.894202	 lr=0.010000
Epoch[003] Batch [1549]/[3760]	Speed: 68.360101 samples/sec	 accuracy=34.951613	 loss=2.891028	 lr=0.010000
Epoch[003] Batch [1599]/[3760]	Speed: 65.913795 samples/sec	 accuracy=34.938477	 loss=2.890203	 lr=0.010000
Epoch[003] Batch [1649]/[3760]	Speed: 67.114235 samples/sec	 accuracy=35.002841	 loss=2.887793	 lr=0.010000
Epoch[003] Batch [1699]/[3760]	Speed: 68.909336 samples/sec	 accuracy=34.962316	 loss=2.890193	 lr=0.010000
Epoch[003] Batch [1749]/[3760]	Speed: 65.531644 samples/sec	 accuracy=34.958929	 loss=2.889867	 lr=0.010000
Epoch[003] Batch [1799]/[3760]	Speed: 66.489141 samples/sec	 accuracy=34.984375	 loss=2.888936	 lr=0.010000
Epoch[003] Batch [1849]/[3760]	Speed: 67.006497 samples/sec	 accuracy=35.011824	 loss=2.887790	 lr=0.010000
Epoch[003] Batch [1899]/[3760]	Speed: 65.762898 samples/sec	 accuracy=35.022204	 loss=2.886009	 lr=0.010000
Epoch[003] Batch [1949]/[3760]	Speed: 67.784990 samples/sec	 accuracy=35.005609	 loss=2.887198	 lr=0.010000
Epoch[003] Batch [1999]/[3760]	Speed: 68.963690 samples/sec	 accuracy=34.997656	 loss=2.888130	 lr=0.010000
Epoch[003] Batch [2049]/[3760]	Speed: 63.990398 samples/sec	 accuracy=34.996951	 loss=2.888389	 lr=0.010000
Epoch[003] Batch [2099]/[3760]	Speed: 68.047505 samples/sec	 accuracy=35.019345	 loss=2.887083	 lr=0.010000
Epoch[003] Batch [2149]/[3760]	Speed: 66.021305 samples/sec	 accuracy=35.048692	 loss=2.885582	 lr=0.010000
Epoch[003] Batch [2199]/[3760]	Speed: 68.637965 samples/sec	 accuracy=35.044744	 loss=2.884957	 lr=0.010000
Epoch[003] Batch [2249]/[3760]	Speed: 66.304333 samples/sec	 accuracy=35.054167	 loss=2.884784	 lr=0.010000
Epoch[003] Batch [2299]/[3760]	Speed: 64.691047 samples/sec	 accuracy=35.005435	 loss=2.886415	 lr=0.010000
Epoch[003] Batch [2349]/[3760]	Speed: 67.288074 samples/sec	 accuracy=35.008644	 loss=2.886074	 lr=0.010000
Epoch[003] Batch [2399]/[3760]	Speed: 67.612359 samples/sec	 accuracy=35.033854	 loss=2.884606	 lr=0.010000
Epoch[003] Batch [2449]/[3760]	Speed: 64.314327 samples/sec	 accuracy=35.061224	 loss=2.883852	 lr=0.010000
Epoch[003] Batch [2499]/[3760]	Speed: 68.281222 samples/sec	 accuracy=35.078750	 loss=2.882780	 lr=0.010000
Epoch[003] Batch [2549]/[3760]	Speed: 65.643592 samples/sec	 accuracy=35.096814	 loss=2.881966	 lr=0.010000
Epoch[003] Batch [2599]/[3760]	Speed: 66.737326 samples/sec	 accuracy=35.107572	 loss=2.882086	 lr=0.010000
Epoch[003] Batch [2649]/[3760]	Speed: 64.570922 samples/sec	 accuracy=35.140920	 loss=2.880460	 lr=0.010000
Epoch[003] Batch [2699]/[3760]	Speed: 67.264470 samples/sec	 accuracy=35.125000	 loss=2.880508	 lr=0.010000
Epoch[003] Batch [2749]/[3760]	Speed: 67.808269 samples/sec	 accuracy=35.153409	 loss=2.879025	 lr=0.010000
Epoch[003] Batch [2799]/[3760]	Speed: 67.639125 samples/sec	 accuracy=35.161830	 loss=2.877891	 lr=0.010000
Epoch[003] Batch [2849]/[3760]	Speed: 64.912059 samples/sec	 accuracy=35.177083	 loss=2.876890	 lr=0.010000
Epoch[003] Batch [2899]/[3760]	Speed: 64.613551 samples/sec	 accuracy=35.191272	 loss=2.875652	 lr=0.010000
Epoch[003] Batch [2949]/[3760]	Speed: 68.172993 samples/sec	 accuracy=35.246292	 loss=2.874346	 lr=0.010000
Epoch[003] Batch [2999]/[3760]	Speed: 67.456190 samples/sec	 accuracy=35.283854	 loss=2.872769	 lr=0.010000
Epoch[003] Batch [3049]/[3760]	Speed: 68.416192 samples/sec	 accuracy=35.291496	 loss=2.871528	 lr=0.010000
Epoch[003] Batch [3099]/[3760]	Speed: 65.641986 samples/sec	 accuracy=35.315020	 loss=2.869849	 lr=0.010000
Epoch[003] Batch [3149]/[3760]	Speed: 66.486087 samples/sec	 accuracy=35.325893	 loss=2.869899	 lr=0.010000
Epoch[003] Batch [3199]/[3760]	Speed: 62.694225 samples/sec	 accuracy=35.319824	 loss=2.869554	 lr=0.010000
Epoch[003] Batch [3249]/[3760]	Speed: 69.669961 samples/sec	 accuracy=35.305769	 loss=2.869749	 lr=0.010000
Epoch[003] Batch [3299]/[3760]	Speed: 68.798389 samples/sec	 accuracy=35.323390	 loss=2.869127	 lr=0.010000
Epoch[003] Batch [3349]/[3760]	Speed: 64.584912 samples/sec	 accuracy=35.317164	 loss=2.869314	 lr=0.010000
Epoch[003] Batch [3399]/[3760]	Speed: 67.936113 samples/sec	 accuracy=35.348805	 loss=2.867931	 lr=0.010000
Epoch[003] Batch [3449]/[3760]	Speed: 65.887888 samples/sec	 accuracy=35.365036	 loss=2.867137	 lr=0.010000
Epoch[003] Batch [3499]/[3760]	Speed: 68.515348 samples/sec	 accuracy=35.389732	 loss=2.865249	 lr=0.010000
Epoch[003] Batch [3549]/[3760]	Speed: 66.412382 samples/sec	 accuracy=35.402289	 loss=2.865078	 lr=0.010000
Epoch[003] Batch [3599]/[3760]	Speed: 64.886126 samples/sec	 accuracy=35.407118	 loss=2.864453	 lr=0.010000
Epoch[003] Batch [3649]/[3760]	Speed: 65.231733 samples/sec	 accuracy=35.431507	 loss=2.863593	 lr=0.010000
Epoch[003] Batch [3699]/[3760]	Speed: 66.370621 samples/sec	 accuracy=35.433699	 loss=2.863507	 lr=0.010000
Epoch[003] Batch [3749]/[3760]	Speed: 75.252458 samples/sec	 accuracy=35.455417	 loss=2.862424	 lr=0.010000
Batch [0049]/[0303]: acc-top1=40.062500 acc-top5=67.375000
Batch [0099]/[0303]: acc-top1=39.765625 acc-top5=67.718750
Batch [0149]/[0303]: acc-top1=39.937500 acc-top5=67.843750
Batch [0199]/[0303]: acc-top1=40.265625 acc-top5=67.898438
Batch [0249]/[0303]: acc-top1=40.275000 acc-top5=67.987500
Batch [0299]/[0303]: acc-top1=40.036458 acc-top5=67.776042
[Epoch 003] training: accuracy=35.460023	 loss=2.862244
[Epoch 003] speed: 66 samples/sec	time cost: 3934.246470
[Epoch 003] validation: acc-top1=40.037129 acc-top5=67.728960 loss=2.737825
Epoch[004] Batch [0049]/[3760]	Speed: 45.604905 samples/sec	 accuracy=38.062500	 loss=2.752838	 lr=0.010000
Epoch[004] Batch [0099]/[3760]	Speed: 68.348354 samples/sec	 accuracy=37.343750	 loss=2.773288	 lr=0.010000
Epoch[004] Batch [0149]/[3760]	Speed: 68.283070 samples/sec	 accuracy=37.437500	 loss=2.761456	 lr=0.010000
Epoch[004] Batch [0199]/[3760]	Speed: 63.828098 samples/sec	 accuracy=37.406250	 loss=2.748386	 lr=0.010000
Epoch[004] Batch [0249]/[3760]	Speed: 67.838650 samples/sec	 accuracy=37.487500	 loss=2.738487	 lr=0.010000
Epoch[004] Batch [0299]/[3760]	Speed: 69.233679 samples/sec	 accuracy=37.421875	 loss=2.744995	 lr=0.010000
Epoch[004] Batch [0349]/[3760]	Speed: 61.617928 samples/sec	 accuracy=37.482143	 loss=2.741452	 lr=0.010000
Epoch[004] Batch [0399]/[3760]	Speed: 69.561347 samples/sec	 accuracy=37.476562	 loss=2.749278	 lr=0.010000
Epoch[004] Batch [0449]/[3760]	Speed: 67.522191 samples/sec	 accuracy=37.482639	 loss=2.748664	 lr=0.010000
Epoch[004] Batch [0499]/[3760]	Speed: 68.217634 samples/sec	 accuracy=37.615625	 loss=2.743932	 lr=0.010000
Epoch[004] Batch [0549]/[3760]	Speed: 64.620368 samples/sec	 accuracy=37.519886	 loss=2.748096	 lr=0.010000
Epoch[004] Batch [0599]/[3760]	Speed: 63.764836 samples/sec	 accuracy=37.635417	 loss=2.745234	 lr=0.010000
Epoch[004] Batch [0649]/[3760]	Speed: 70.729032 samples/sec	 accuracy=37.692308	 loss=2.745265	 lr=0.010000
Epoch[004] Batch [0699]/[3760]	Speed: 67.974168 samples/sec	 accuracy=37.689732	 loss=2.745815	 lr=0.010000
Epoch[004] Batch [0749]/[3760]	Speed: 64.830221 samples/sec	 accuracy=37.722917	 loss=2.743016	 lr=0.010000
Epoch[004] Batch [0799]/[3760]	Speed: 66.882091 samples/sec	 accuracy=37.751953	 loss=2.742515	 lr=0.010000
Epoch[004] Batch [0849]/[3760]	Speed: 65.659072 samples/sec	 accuracy=37.729779	 loss=2.743149	 lr=0.010000
Epoch[004] Batch [0899]/[3760]	Speed: 67.442108 samples/sec	 accuracy=37.730903	 loss=2.741507	 lr=0.010000
Epoch[004] Batch [0949]/[3760]	Speed: 68.374081 samples/sec	 accuracy=37.712171	 loss=2.741120	 lr=0.010000
Epoch[004] Batch [0999]/[3760]	Speed: 65.746012 samples/sec	 accuracy=37.748438	 loss=2.739880	 lr=0.010000
Epoch[004] Batch [1049]/[3760]	Speed: 66.315907 samples/sec	 accuracy=37.754464	 loss=2.740146	 lr=0.010000
Epoch[004] Batch [1099]/[3760]	Speed: 68.317611 samples/sec	 accuracy=37.808239	 loss=2.737835	 lr=0.010000
Epoch[004] Batch [1149]/[3760]	Speed: 66.267288 samples/sec	 accuracy=37.808424	 loss=2.739059	 lr=0.010000
Epoch[004] Batch [1199]/[3760]	Speed: 66.285717 samples/sec	 accuracy=37.774740	 loss=2.740813	 lr=0.010000
Epoch[004] Batch [1249]/[3760]	Speed: 66.333053 samples/sec	 accuracy=37.747500	 loss=2.741964	 lr=0.010000
Epoch[004] Batch [1299]/[3760]	Speed: 67.272737 samples/sec	 accuracy=37.799279	 loss=2.740979	 lr=0.010000
Epoch[004] Batch [1349]/[3760]	Speed: 66.723819 samples/sec	 accuracy=37.778935	 loss=2.741644	 lr=0.010000
Epoch[004] Batch [1399]/[3760]	Speed: 68.354172 samples/sec	 accuracy=37.803571	 loss=2.742378	 lr=0.010000
Epoch[004] Batch [1449]/[3760]	Speed: 63.612955 samples/sec	 accuracy=37.804957	 loss=2.741995	 lr=0.010000
Epoch[004] Batch [1499]/[3760]	Speed: 69.296075 samples/sec	 accuracy=37.808333	 loss=2.741325	 lr=0.010000
Epoch[004] Batch [1549]/[3760]	Speed: 63.156928 samples/sec	 accuracy=37.836694	 loss=2.740982	 lr=0.010000
Epoch[004] Batch [1599]/[3760]	Speed: 70.406912 samples/sec	 accuracy=37.826172	 loss=2.740468	 lr=0.010000
Epoch[004] Batch [1649]/[3760]	Speed: 67.146439 samples/sec	 accuracy=37.778409	 loss=2.741328	 lr=0.010000
Epoch[004] Batch [1699]/[3760]	Speed: 66.151966 samples/sec	 accuracy=37.776654	 loss=2.740408	 lr=0.010000
Epoch[004] Batch [1749]/[3760]	Speed: 66.591786 samples/sec	 accuracy=37.771429	 loss=2.740815	 lr=0.010000
Epoch[004] Batch [1799]/[3760]	Speed: 65.156272 samples/sec	 accuracy=37.789931	 loss=2.740104	 lr=0.010000
Epoch[004] Batch [1849]/[3760]	Speed: 68.318768 samples/sec	 accuracy=37.768581	 loss=2.741112	 lr=0.010000
Epoch[004] Batch [1899]/[3760]	Speed: 67.467135 samples/sec	 accuracy=37.785362	 loss=2.739425	 lr=0.010000
Epoch[004] Batch [1949]/[3760]	Speed: 63.662964 samples/sec	 accuracy=37.772436	 loss=2.739990	 lr=0.010000
Epoch[004] Batch [1999]/[3760]	Speed: 69.432417 samples/sec	 accuracy=37.850000	 loss=2.738210	 lr=0.010000
Epoch[004] Batch [2049]/[3760]	Speed: 67.470200 samples/sec	 accuracy=37.862805	 loss=2.738087	 lr=0.010000
Epoch[004] Batch [2099]/[3760]	Speed: 66.235180 samples/sec	 accuracy=37.877976	 loss=2.737267	 lr=0.010000
Epoch[004] Batch [2149]/[3760]	Speed: 66.662711 samples/sec	 accuracy=37.882267	 loss=2.736663	 lr=0.010000
Epoch[004] Batch [2199]/[3760]	Speed: 66.341835 samples/sec	 accuracy=37.900568	 loss=2.735860	 lr=0.010000
Epoch[004] Batch [2249]/[3760]	Speed: 66.703082 samples/sec	 accuracy=37.906250	 loss=2.735863	 lr=0.010000
Epoch[004] Batch [2299]/[3760]	Speed: 66.774214 samples/sec	 accuracy=37.915761	 loss=2.736854	 lr=0.010000
Epoch[004] Batch [2349]/[3760]	Speed: 68.434118 samples/sec	 accuracy=37.938830	 loss=2.735919	 lr=0.010000
Epoch[004] Batch [2399]/[3760]	Speed: 66.449636 samples/sec	 accuracy=37.952474	 loss=2.736021	 lr=0.010000
Epoch[004] Batch [2449]/[3760]	Speed: 65.325858 samples/sec	 accuracy=37.965561	 loss=2.734772	 lr=0.010000
Epoch[004] Batch [2499]/[3760]	Speed: 69.488278 samples/sec	 accuracy=37.983750	 loss=2.733165	 lr=0.010000
Epoch[004] Batch [2549]/[3760]	Speed: 67.447152 samples/sec	 accuracy=38.001838	 loss=2.732362	 lr=0.010000
Epoch[004] Batch [2599]/[3760]	Speed: 65.406129 samples/sec	 accuracy=38.006611	 loss=2.731269	 lr=0.010000
Epoch[004] Batch [2649]/[3760]	Speed: 67.465123 samples/sec	 accuracy=38.026533	 loss=2.730094	 lr=0.010000
Epoch[004] Batch [2699]/[3760]	Speed: 66.613624 samples/sec	 accuracy=38.028356	 loss=2.730925	 lr=0.010000
Epoch[004] Batch [2749]/[3760]	Speed: 68.571831 samples/sec	 accuracy=38.031250	 loss=2.730235	 lr=0.010000
Epoch[004] Batch [2799]/[3760]	Speed: 65.829025 samples/sec	 accuracy=38.052455	 loss=2.729189	 lr=0.010000
Epoch[004] Batch [2849]/[3760]	Speed: 68.009934 samples/sec	 accuracy=38.050439	 loss=2.729228	 lr=0.010000
Epoch[004] Batch [2899]/[3760]	Speed: 64.723590 samples/sec	 accuracy=38.057651	 loss=2.728369	 lr=0.010000
Epoch[004] Batch [2949]/[3760]	Speed: 68.351867 samples/sec	 accuracy=38.051907	 loss=2.728979	 lr=0.010000
Epoch[004] Batch [2999]/[3760]	Speed: 66.835096 samples/sec	 accuracy=38.092708	 loss=2.728047	 lr=0.010000
Epoch[004] Batch [3049]/[3760]	Speed: 67.529891 samples/sec	 accuracy=38.111680	 loss=2.726944	 lr=0.010000
Epoch[004] Batch [3099]/[3760]	Speed: 64.708236 samples/sec	 accuracy=38.122984	 loss=2.725725	 lr=0.010000
Epoch[004] Batch [3149]/[3760]	Speed: 68.212890 samples/sec	 accuracy=38.123512	 loss=2.725339	 lr=0.010000
Epoch[004] Batch [3199]/[3760]	Speed: 67.194243 samples/sec	 accuracy=38.132812	 loss=2.725269	 lr=0.010000
Epoch[004] Batch [3249]/[3760]	Speed: 66.261554 samples/sec	 accuracy=38.151923	 loss=2.724074	 lr=0.010000
Epoch[004] Batch [3299]/[3760]	Speed: 65.905913 samples/sec	 accuracy=38.157670	 loss=2.723001	 lr=0.010000
Epoch[004] Batch [3349]/[3760]	Speed: 68.298437 samples/sec	 accuracy=38.157649	 loss=2.722683	 lr=0.010000
Epoch[004] Batch [3399]/[3760]	Speed: 63.111894 samples/sec	 accuracy=38.166820	 loss=2.722106	 lr=0.010000
Epoch[004] Batch [3449]/[3760]	Speed: 69.336427 samples/sec	 accuracy=38.194293	 loss=2.721591	 lr=0.010000
Epoch[004] Batch [3499]/[3760]	Speed: 68.904058 samples/sec	 accuracy=38.202232	 loss=2.721190	 lr=0.010000
Epoch[004] Batch [3549]/[3760]	Speed: 63.978474 samples/sec	 accuracy=38.220951	 loss=2.720492	 lr=0.010000
Epoch[004] Batch [3599]/[3760]	Speed: 67.823497 samples/sec	 accuracy=38.230035	 loss=2.719751	 lr=0.010000
Epoch[004] Batch [3649]/[3760]	Speed: 66.721237 samples/sec	 accuracy=38.244435	 loss=2.718908	 lr=0.010000
Epoch[004] Batch [3699]/[3760]	Speed: 66.124380 samples/sec	 accuracy=38.249578	 loss=2.719200	 lr=0.010000
Epoch[004] Batch [3749]/[3760]	Speed: 75.481506 samples/sec	 accuracy=38.252500	 loss=2.718796	 lr=0.010000
Batch [0049]/[0303]: acc-top1=42.031250 acc-top5=68.656250
Batch [0099]/[0303]: acc-top1=41.187500 acc-top5=68.406250
Batch [0149]/[0303]: acc-top1=41.562500 acc-top5=69.041667
Batch [0199]/[0303]: acc-top1=41.929688 acc-top5=69.445312
Batch [0249]/[0303]: acc-top1=41.993750 acc-top5=69.443750
Batch [0299]/[0303]: acc-top1=41.843750 acc-top5=69.333333
[Epoch 004] training: accuracy=38.246343	 loss=2.718849
[Epoch 004] speed: 66 samples/sec	time cost: 3926.086306
[Epoch 004] validation: acc-top1=41.836840 acc-top5=69.286304 loss=2.633419
Epoch[005] Batch [0049]/[3759]	Speed: 47.447364 samples/sec	 accuracy=39.593750	 loss=2.627201	 lr=0.010000
Epoch[005] Batch [0099]/[3759]	Speed: 64.449671 samples/sec	 accuracy=39.859375	 loss=2.611551	 lr=0.010000
Epoch[005] Batch [0149]/[3759]	Speed: 68.121023 samples/sec	 accuracy=40.062500	 loss=2.613328	 lr=0.010000
Epoch[005] Batch [0199]/[3759]	Speed: 67.010467 samples/sec	 accuracy=39.828125	 loss=2.619670	 lr=0.010000
Epoch[005] Batch [0249]/[3759]	Speed: 67.710618 samples/sec	 accuracy=40.056250	 loss=2.614114	 lr=0.010000
Epoch[005] Batch [0299]/[3759]	Speed: 66.933373 samples/sec	 accuracy=39.843750	 loss=2.617498	 lr=0.010000
Epoch[005] Batch [0349]/[3759]	Speed: 67.206302 samples/sec	 accuracy=39.825893	 loss=2.623155	 lr=0.010000
Epoch[005] Batch [0399]/[3759]	Speed: 65.947098 samples/sec	 accuracy=39.546875	 loss=2.636131	 lr=0.010000
Epoch[005] Batch [0449]/[3759]	Speed: 67.130877 samples/sec	 accuracy=39.541667	 loss=2.636889	 lr=0.010000
Epoch[005] Batch [0499]/[3759]	Speed: 65.764649 samples/sec	 accuracy=39.765625	 loss=2.630947	 lr=0.010000
Epoch[005] Batch [0549]/[3759]	Speed: 67.994959 samples/sec	 accuracy=39.741477	 loss=2.629494	 lr=0.010000
Epoch[005] Batch [0599]/[3759]	Speed: 65.795606 samples/sec	 accuracy=39.713542	 loss=2.630470	 lr=0.010000
Epoch[005] Batch [0649]/[3759]	Speed: 67.799534 samples/sec	 accuracy=39.733173	 loss=2.628326	 lr=0.010000
Epoch[005] Batch [0699]/[3759]	Speed: 64.319928 samples/sec	 accuracy=39.823661	 loss=2.625902	 lr=0.010000
Epoch[005] Batch [0749]/[3759]	Speed: 67.473658 samples/sec	 accuracy=39.960417	 loss=2.621372	 lr=0.010000
Epoch[005] Batch [0799]/[3759]	Speed: 68.893436 samples/sec	 accuracy=39.962891	 loss=2.624035	 lr=0.010000
Epoch[005] Batch [0849]/[3759]	Speed: 63.472468 samples/sec	 accuracy=39.930147	 loss=2.623533	 lr=0.010000
Epoch[005] Batch [0899]/[3759]	Speed: 69.257690 samples/sec	 accuracy=39.927083	 loss=2.622869	 lr=0.010000
Epoch[005] Batch [0949]/[3759]	Speed: 65.627746 samples/sec	 accuracy=39.947368	 loss=2.623057	 lr=0.010000
Epoch[005] Batch [0999]/[3759]	Speed: 68.927384 samples/sec	 accuracy=40.000000	 loss=2.618864	 lr=0.010000
Epoch[005] Batch [1049]/[3759]	Speed: 66.580282 samples/sec	 accuracy=40.037202	 loss=2.618401	 lr=0.010000
Epoch[005] Batch [1099]/[3759]	Speed: 66.337652 samples/sec	 accuracy=39.990057	 loss=2.622162	 lr=0.010000
Epoch[005] Batch [1149]/[3759]	Speed: 67.976636 samples/sec	 accuracy=39.927989	 loss=2.623062	 lr=0.010000
Epoch[005] Batch [1199]/[3759]	Speed: 64.947046 samples/sec	 accuracy=39.924479	 loss=2.623108	 lr=0.010000
Epoch[005] Batch [1249]/[3759]	Speed: 67.849995 samples/sec	 accuracy=39.863750	 loss=2.625814	 lr=0.010000
Epoch[005] Batch [1299]/[3759]	Speed: 64.458682 samples/sec	 accuracy=39.861779	 loss=2.624932	 lr=0.010000
Epoch[005] Batch [1349]/[3759]	Speed: 68.805031 samples/sec	 accuracy=39.862269	 loss=2.624916	 lr=0.010000
Epoch[005] Batch [1399]/[3759]	Speed: 68.128691 samples/sec	 accuracy=39.910714	 loss=2.622852	 lr=0.010000
Epoch[005] Batch [1449]/[3759]	Speed: 65.354754 samples/sec	 accuracy=39.915948	 loss=2.622749	 lr=0.010000
Epoch[005] Batch [1499]/[3759]	Speed: 67.952699 samples/sec	 accuracy=39.918750	 loss=2.622801	 lr=0.010000
Epoch[005] Batch [1549]/[3759]	Speed: 63.651902 samples/sec	 accuracy=39.904234	 loss=2.621712	 lr=0.010000
Epoch[005] Batch [1599]/[3759]	Speed: 69.666051 samples/sec	 accuracy=39.911133	 loss=2.620504	 lr=0.010000
Epoch[005] Batch [1649]/[3759]	Speed: 67.930822 samples/sec	 accuracy=39.940341	 loss=2.621380	 lr=0.010000
Epoch[005] Batch [1699]/[3759]	Speed: 66.290666 samples/sec	 accuracy=39.943934	 loss=2.622337	 lr=0.010000
Epoch[005] Batch [1749]/[3759]	Speed: 66.712760 samples/sec	 accuracy=39.983036	 loss=2.620524	 lr=0.010000
Epoch[005] Batch [1799]/[3759]	Speed: 64.432621 samples/sec	 accuracy=40.023438	 loss=2.618486	 lr=0.010000
Epoch[005] Batch [1849]/[3759]	Speed: 68.941058 samples/sec	 accuracy=40.024493	 loss=2.618288	 lr=0.010000
Epoch[005] Batch [1899]/[3759]	Speed: 64.742064 samples/sec	 accuracy=40.041941	 loss=2.617584	 lr=0.010000
Epoch[005] Batch [1949]/[3759]	Speed: 68.420437 samples/sec	 accuracy=40.036859	 loss=2.616950	 lr=0.010000
Epoch[005] Batch [1999]/[3759]	Speed: 66.656917 samples/sec	 accuracy=40.073437	 loss=2.614877	 lr=0.010000
Epoch[005] Batch [2049]/[3759]	Speed: 68.349709 samples/sec	 accuracy=40.128049	 loss=2.612445	 lr=0.010000
Epoch[005] Batch [2099]/[3759]	Speed: 66.432058 samples/sec	 accuracy=40.115327	 loss=2.613577	 lr=0.010000
Epoch[005] Batch [2149]/[3759]	Speed: 67.318426 samples/sec	 accuracy=40.132994	 loss=2.612746	 lr=0.010000
Epoch[005] Batch [2199]/[3759]	Speed: 67.169755 samples/sec	 accuracy=40.137784	 loss=2.612693	 lr=0.010000
Epoch[005] Batch [2249]/[3759]	Speed: 67.073552 samples/sec	 accuracy=40.175000	 loss=2.611790	 lr=0.010000
Epoch[005] Batch [2299]/[3759]	Speed: 63.938787 samples/sec	 accuracy=40.192255	 loss=2.611615	 lr=0.010000
Epoch[005] Batch [2349]/[3759]	Speed: 69.035574 samples/sec	 accuracy=40.191489	 loss=2.611635	 lr=0.010000
Epoch[005] Batch [2399]/[3759]	Speed: 68.167857 samples/sec	 accuracy=40.224609	 loss=2.610287	 lr=0.010000
Epoch[005] Batch [2449]/[3759]	Speed: 67.363099 samples/sec	 accuracy=40.186224	 loss=2.611899	 lr=0.010000
Epoch[005] Batch [2499]/[3759]	Speed: 64.286616 samples/sec	 accuracy=40.187500	 loss=2.611557	 lr=0.010000
Epoch[005] Batch [2549]/[3759]	Speed: 68.160146 samples/sec	 accuracy=40.209559	 loss=2.610915	 lr=0.010000
Epoch[005] Batch [2599]/[3759]	Speed: 65.883497 samples/sec	 accuracy=40.173077	 loss=2.612659	 lr=0.010000
Epoch[005] Batch [2649]/[3759]	Speed: 68.447936 samples/sec	 accuracy=40.166274	 loss=2.612042	 lr=0.010000
Epoch[005] Batch [2699]/[3759]	Speed: 67.282731 samples/sec	 accuracy=40.197338	 loss=2.610006	 lr=0.010000
Epoch[005] Batch [2749]/[3759]	Speed: 65.974225 samples/sec	 accuracy=40.209091	 loss=2.609917	 lr=0.010000
Epoch[005] Batch [2799]/[3759]	Speed: 66.918758 samples/sec	 accuracy=40.217634	 loss=2.609865	 lr=0.010000
Epoch[005] Batch [2849]/[3759]	Speed: 67.297550 samples/sec	 accuracy=40.213268	 loss=2.609970	 lr=0.010000
Epoch[005] Batch [2899]/[3759]	Speed: 66.183513 samples/sec	 accuracy=40.206897	 loss=2.610258	 lr=0.010000
Epoch[005] Batch [2949]/[3759]	Speed: 68.277192 samples/sec	 accuracy=40.217691	 loss=2.609690	 lr=0.010000
Epoch[005] Batch [2999]/[3759]	Speed: 63.980066 samples/sec	 accuracy=40.247396	 loss=2.608730	 lr=0.010000
Epoch[005] Batch [3049]/[3759]	Speed: 68.995614 samples/sec	 accuracy=40.233607	 loss=2.609130	 lr=0.010000
Epoch[005] Batch [3099]/[3759]	Speed: 67.447673 samples/sec	 accuracy=40.256552	 loss=2.608295	 lr=0.010000
Epoch[005] Batch [3149]/[3759]	Speed: 67.240312 samples/sec	 accuracy=40.287698	 loss=2.607301	 lr=0.010000
Epoch[005] Batch [3199]/[3759]	Speed: 65.432069 samples/sec	 accuracy=40.303711	 loss=2.606219	 lr=0.010000
Epoch[005] Batch [3249]/[3759]	Speed: 68.232063 samples/sec	 accuracy=40.323077	 loss=2.605581	 lr=0.010000
Epoch[005] Batch [3299]/[3759]	Speed: 65.574650 samples/sec	 accuracy=40.323864	 loss=2.605323	 lr=0.010000
Epoch[005] Batch [3349]/[3759]	Speed: 69.565819 samples/sec	 accuracy=40.342351	 loss=2.604264	 lr=0.010000
Epoch[005] Batch [3399]/[3759]	Speed: 64.680082 samples/sec	 accuracy=40.365349	 loss=2.603194	 lr=0.010000
Epoch[005] Batch [3449]/[3759]	Speed: 66.951486 samples/sec	 accuracy=40.371830	 loss=2.603109	 lr=0.010000
Epoch[005] Batch [3499]/[3759]	Speed: 68.253832 samples/sec	 accuracy=40.365625	 loss=2.603274	 lr=0.010000
Epoch[005] Batch [3549]/[3759]	Speed: 66.928292 samples/sec	 accuracy=40.380722	 loss=2.602682	 lr=0.010000
Epoch[005] Batch [3599]/[3759]	Speed: 65.650456 samples/sec	 accuracy=40.398872	 loss=2.602106	 lr=0.010000
Epoch[005] Batch [3649]/[3759]	Speed: 67.612184 samples/sec	 accuracy=40.388699	 loss=2.602780	 lr=0.010000
Epoch[005] Batch [3699]/[3759]	Speed: 67.931865 samples/sec	 accuracy=40.404561	 loss=2.601946	 lr=0.010000
Epoch[005] Batch [3749]/[3759]	Speed: 73.908202 samples/sec	 accuracy=40.422083	 loss=2.601571	 lr=0.010000
Batch [0049]/[0303]: acc-top1=43.875000 acc-top5=70.968750
Batch [0099]/[0303]: acc-top1=43.031250 acc-top5=70.500000
Batch [0149]/[0303]: acc-top1=43.406250 acc-top5=70.791667
Batch [0199]/[0303]: acc-top1=43.726562 acc-top5=70.906250
Batch [0249]/[0303]: acc-top1=43.581250 acc-top5=71.018750
Batch [0299]/[0303]: acc-top1=43.411458 acc-top5=70.864583
[Epoch 005] training: accuracy=40.420075	 loss=2.601741
[Epoch 005] speed: 66 samples/sec	time cost: 3919.177756
[Epoch 005] validation: acc-top1=43.414810 acc-top5=70.812706 loss=2.549411
Epoch[006] Batch [0049]/[3760]	Speed: 46.731260 samples/sec	 accuracy=42.531250	 loss=2.533641	 lr=0.010000
Epoch[006] Batch [0099]/[3760]	Speed: 66.661610 samples/sec	 accuracy=42.750000	 loss=2.528290	 lr=0.010000
Epoch[006] Batch [0149]/[3760]	Speed: 66.433538 samples/sec	 accuracy=42.791667	 loss=2.518459	 lr=0.010000
Epoch[006] Batch [0199]/[3760]	Speed: 65.867980 samples/sec	 accuracy=42.921875	 loss=2.507514	 lr=0.010000
Epoch[006] Batch [0249]/[3760]	Speed: 66.395850 samples/sec	 accuracy=42.500000	 loss=2.516499	 lr=0.010000
Epoch[006] Batch [0299]/[3760]	Speed: 69.349149 samples/sec	 accuracy=42.479167	 loss=2.515221	 lr=0.010000
Epoch[006] Batch [0349]/[3760]	Speed: 66.622635 samples/sec	 accuracy=42.473214	 loss=2.510689	 lr=0.010000
Epoch[006] Batch [0399]/[3760]	Speed: 66.718759 samples/sec	 accuracy=42.316406	 loss=2.514088	 lr=0.010000
Epoch[006] Batch [0449]/[3760]	Speed: 67.691178 samples/sec	 accuracy=42.131944	 loss=2.517575	 lr=0.010000
Epoch[006] Batch [0499]/[3760]	Speed: 64.360161 samples/sec	 accuracy=42.015625	 loss=2.521748	 lr=0.010000
Epoch[006] Batch [0549]/[3760]	Speed: 68.878357 samples/sec	 accuracy=42.062500	 loss=2.523118	 lr=0.010000
Epoch[006] Batch [0599]/[3760]	Speed: 67.267691 samples/sec	 accuracy=42.114583	 loss=2.524000	 lr=0.010000
Epoch[006] Batch [0649]/[3760]	Speed: 65.931798 samples/sec	 accuracy=42.153846	 loss=2.520859	 lr=0.010000
Epoch[006] Batch [0699]/[3760]	Speed: 65.818354 samples/sec	 accuracy=42.142857	 loss=2.525026	 lr=0.010000
Epoch[006] Batch [0749]/[3760]	Speed: 68.617615 samples/sec	 accuracy=42.214583	 loss=2.522591	 lr=0.010000
Epoch[006] Batch [0799]/[3760]	Speed: 66.311599 samples/sec	 accuracy=42.173828	 loss=2.524938	 lr=0.010000
Epoch[006] Batch [0849]/[3760]	Speed: 68.564114 samples/sec	 accuracy=42.042279	 loss=2.527779	 lr=0.010000
Epoch[006] Batch [0899]/[3760]	Speed: 64.726645 samples/sec	 accuracy=41.984375	 loss=2.530245	 lr=0.010000
Epoch[006] Batch [0949]/[3760]	Speed: 66.943278 samples/sec	 accuracy=42.018092	 loss=2.529724	 lr=0.010000
Epoch[006] Batch [0999]/[3760]	Speed: 69.690791 samples/sec	 accuracy=42.021875	 loss=2.531057	 lr=0.010000
Epoch[006] Batch [1049]/[3760]	Speed: 66.234879 samples/sec	 accuracy=42.007440	 loss=2.530054	 lr=0.010000
Epoch[006] Batch [1099]/[3760]	Speed: 67.451129 samples/sec	 accuracy=41.968750	 loss=2.530360	 lr=0.010000
Epoch[006] Batch [1149]/[3760]	Speed: 64.037604 samples/sec	 accuracy=42.050272	 loss=2.526484	 lr=0.010000
Epoch[006] Batch [1199]/[3760]	Speed: 68.669147 samples/sec	 accuracy=42.048177	 loss=2.528253	 lr=0.010000
Epoch[006] Batch [1249]/[3760]	Speed: 68.763980 samples/sec	 accuracy=42.038750	 loss=2.527527	 lr=0.010000
Epoch[006] Batch [1299]/[3760]	Speed: 65.916582 samples/sec	 accuracy=42.021635	 loss=2.529101	 lr=0.010000
Epoch[006] Batch [1349]/[3760]	Speed: 65.419772 samples/sec	 accuracy=42.009259	 loss=2.530216	 lr=0.010000
Epoch[006] Batch [1399]/[3760]	Speed: 67.954691 samples/sec	 accuracy=41.966518	 loss=2.531455	 lr=0.010000
Epoch[006] Batch [1449]/[3760]	Speed: 68.477439 samples/sec	 accuracy=41.926724	 loss=2.532652	 lr=0.010000
Epoch[006] Batch [1499]/[3760]	Speed: 65.860616 samples/sec	 accuracy=41.902083	 loss=2.535222	 lr=0.010000
Epoch[006] Batch [1549]/[3760]	Speed: 67.404230 samples/sec	 accuracy=41.935484	 loss=2.533775	 lr=0.010000
Epoch[006] Batch [1599]/[3760]	Speed: 66.302217 samples/sec	 accuracy=41.955078	 loss=2.532292	 lr=0.010000
Epoch[006] Batch [1649]/[3760]	Speed: 66.613963 samples/sec	 accuracy=41.998106	 loss=2.532202	 lr=0.010000
Epoch[006] Batch [1699]/[3760]	Speed: 67.827759 samples/sec	 accuracy=41.980699	 loss=2.532167	 lr=0.010000
Epoch[006] Batch [1749]/[3760]	Speed: 68.389014 samples/sec	 accuracy=42.016071	 loss=2.529581	 lr=0.010000
Epoch[006] Batch [1799]/[3760]	Speed: 65.282435 samples/sec	 accuracy=42.035590	 loss=2.528378	 lr=0.010000
Epoch[006] Batch [1849]/[3760]	Speed: 67.499918 samples/sec	 accuracy=42.024493	 loss=2.528993	 lr=0.010000
Epoch[006] Batch [1899]/[3760]	Speed: 67.776230 samples/sec	 accuracy=42.038651	 loss=2.528305	 lr=0.010000
Epoch[006] Batch [1949]/[3760]	Speed: 67.324125 samples/sec	 accuracy=41.992788	 loss=2.528314	 lr=0.010000
Epoch[006] Batch [1999]/[3760]	Speed: 65.996119 samples/sec	 accuracy=42.038281	 loss=2.526538	 lr=0.010000
Epoch[006] Batch [2049]/[3760]	Speed: 68.258254 samples/sec	 accuracy=42.041159	 loss=2.527072	 lr=0.010000
Epoch[006] Batch [2099]/[3760]	Speed: 65.278749 samples/sec	 accuracy=42.033482	 loss=2.527257	 lr=0.010000
Epoch[006] Batch [2149]/[3760]	Speed: 68.495634 samples/sec	 accuracy=42.034157	 loss=2.526675	 lr=0.010000
Epoch[006] Batch [2199]/[3760]	Speed: 67.387675 samples/sec	 accuracy=42.064631	 loss=2.525879	 lr=0.010000
Epoch[006] Batch [2249]/[3760]	Speed: 65.779864 samples/sec	 accuracy=42.055556	 loss=2.526514	 lr=0.010000
Epoch[006] Batch [2299]/[3760]	Speed: 67.091136 samples/sec	 accuracy=42.037364	 loss=2.526497	 lr=0.010000
Epoch[006] Batch [2349]/[3760]	Speed: 68.280431 samples/sec	 accuracy=42.067819	 loss=2.525385	 lr=0.010000
Epoch[006] Batch [2399]/[3760]	Speed: 68.394625 samples/sec	 accuracy=42.051432	 loss=2.526028	 lr=0.010000
Epoch[006] Batch [2449]/[3760]	Speed: 65.749165 samples/sec	 accuracy=42.035714	 loss=2.526900	 lr=0.010000
Epoch[006] Batch [2499]/[3760]	Speed: 65.698049 samples/sec	 accuracy=42.026250	 loss=2.527009	 lr=0.010000
Epoch[006] Batch [2549]/[3760]	Speed: 67.844728 samples/sec	 accuracy=42.007353	 loss=2.527759	 lr=0.010000
Epoch[006] Batch [2599]/[3760]	Speed: 68.699922 samples/sec	 accuracy=42.018029	 loss=2.527188	 lr=0.010000
Epoch[006] Batch [2649]/[3760]	Speed: 66.744395 samples/sec	 accuracy=42.004127	 loss=2.527491	 lr=0.010000
Epoch[006] Batch [2699]/[3760]	Speed: 66.726362 samples/sec	 accuracy=42.011574	 loss=2.526602	 lr=0.010000
Epoch[006] Batch [2749]/[3760]	Speed: 67.627339 samples/sec	 accuracy=42.005114	 loss=2.526266	 lr=0.010000
Epoch[006] Batch [2799]/[3760]	Speed: 65.642612 samples/sec	 accuracy=41.986049	 loss=2.527165	 lr=0.010000
Epoch[006] Batch [2849]/[3760]	Speed: 66.626301 samples/sec	 accuracy=41.955044	 loss=2.528693	 lr=0.010000
Epoch[006] Batch [2899]/[3760]	Speed: 69.621749 samples/sec	 accuracy=41.960668	 loss=2.528716	 lr=0.010000
Epoch[006] Batch [2949]/[3760]	Speed: 67.356584 samples/sec	 accuracy=41.971928	 loss=2.528290	 lr=0.010000
Epoch[006] Batch [2999]/[3760]	Speed: 65.382846 samples/sec	 accuracy=41.998438	 loss=2.526829	 lr=0.010000
Epoch[006] Batch [3049]/[3760]	Speed: 68.823742 samples/sec	 accuracy=41.991291	 loss=2.526771	 lr=0.010000
Epoch[006] Batch [3099]/[3760]	Speed: 64.083365 samples/sec	 accuracy=42.014113	 loss=2.525571	 lr=0.010000
Epoch[006] Batch [3149]/[3760]	Speed: 68.522302 samples/sec	 accuracy=42.013393	 loss=2.525031	 lr=0.010000
Epoch[006] Batch [3199]/[3760]	Speed: 67.286687 samples/sec	 accuracy=42.027344	 loss=2.524496	 lr=0.010000
Epoch[006] Batch [3249]/[3760]	Speed: 69.465063 samples/sec	 accuracy=42.030288	 loss=2.525482	 lr=0.010000
Epoch[006] Batch [3299]/[3760]	Speed: 64.308871 samples/sec	 accuracy=42.031250	 loss=2.525839	 lr=0.010000
Epoch[006] Batch [3349]/[3760]	Speed: 68.886654 samples/sec	 accuracy=42.033116	 loss=2.525247	 lr=0.010000
Epoch[006] Batch [3399]/[3760]	Speed: 67.791082 samples/sec	 accuracy=42.048254	 loss=2.524735	 lr=0.010000
Epoch[006] Batch [3449]/[3760]	Speed: 65.096032 samples/sec	 accuracy=42.043478	 loss=2.524741	 lr=0.010000
Epoch[006] Batch [3499]/[3760]	Speed: 67.971887 samples/sec	 accuracy=42.054018	 loss=2.524205	 lr=0.010000
Epoch[006] Batch [3549]/[3760]	Speed: 65.868282 samples/sec	 accuracy=42.073944	 loss=2.523729	 lr=0.010000
Epoch[006] Batch [3599]/[3760]	Speed: 68.119088 samples/sec	 accuracy=42.095052	 loss=2.523656	 lr=0.010000
Epoch[006] Batch [3649]/[3760]	Speed: 66.758146 samples/sec	 accuracy=42.117295	 loss=2.523082	 lr=0.010000
Epoch[006] Batch [3699]/[3760]	Speed: 65.425051 samples/sec	 accuracy=42.118243	 loss=2.522600	 lr=0.010000
Epoch[006] Batch [3749]/[3760]	Speed: 76.538748 samples/sec	 accuracy=42.167083	 loss=2.521095	 lr=0.010000
Batch [0049]/[0303]: acc-top1=46.093750 acc-top5=71.750000
Batch [0099]/[0303]: acc-top1=45.187500 acc-top5=71.437500
Batch [0149]/[0303]: acc-top1=45.333333 acc-top5=71.916667
Batch [0199]/[0303]: acc-top1=45.750000 acc-top5=72.187500
Batch [0249]/[0303]: acc-top1=45.700000 acc-top5=72.331250
Batch [0299]/[0303]: acc-top1=45.416667 acc-top5=72.093750
[Epoch 006] training: accuracy=42.161735	 loss=2.521178
[Epoch 006] speed: 66 samples/sec	time cost: 3912.991524
[Epoch 006] validation: acc-top1=45.425949 acc-top5=72.091584 loss=2.474816
Epoch[007] Batch [0049]/[3760]	Speed: 46.459256 samples/sec	 accuracy=43.875000	 loss=2.413219	 lr=0.010000
Epoch[007] Batch [0099]/[3760]	Speed: 67.552671 samples/sec	 accuracy=42.750000	 loss=2.462813	 lr=0.010000
Epoch[007] Batch [0149]/[3760]	Speed: 65.851546 samples/sec	 accuracy=42.906250	 loss=2.458277	 lr=0.010000
Epoch[007] Batch [0199]/[3760]	Speed: 72.884591 samples/sec	 accuracy=43.226562	 loss=2.450408	 lr=0.010000
Epoch[007] Batch [0249]/[3760]	Speed: 66.906462 samples/sec	 accuracy=43.137500	 loss=2.463715	 lr=0.010000
Epoch[007] Batch [0299]/[3760]	Speed: 67.716380 samples/sec	 accuracy=43.161458	 loss=2.464450	 lr=0.010000
Epoch[007] Batch [0349]/[3760]	Speed: 68.608036 samples/sec	 accuracy=43.058036	 loss=2.461614	 lr=0.010000
Epoch[007] Batch [0399]/[3760]	Speed: 67.489480 samples/sec	 accuracy=43.027344	 loss=2.464980	 lr=0.010000
Epoch[007] Batch [0449]/[3760]	Speed: 67.564702 samples/sec	 accuracy=42.895833	 loss=2.472411	 lr=0.010000
Epoch[007] Batch [0499]/[3760]	Speed: 72.349776 samples/sec	 accuracy=43.093750	 loss=2.460589	 lr=0.010000
Epoch[007] Batch [0549]/[3760]	Speed: 63.786277 samples/sec	 accuracy=43.224432	 loss=2.453613	 lr=0.010000
Epoch[007] Batch [0599]/[3760]	Speed: 72.845657 samples/sec	 accuracy=43.317708	 loss=2.446145	 lr=0.010000
Epoch[007] Batch [0649]/[3760]	Speed: 68.454555 samples/sec	 accuracy=43.367788	 loss=2.445953	 lr=0.010000
Epoch[007] Batch [0699]/[3760]	Speed: 67.318747 samples/sec	 accuracy=43.354911	 loss=2.448184	 lr=0.010000
Epoch[007] Batch [0749]/[3760]	Speed: 70.282049 samples/sec	 accuracy=43.337500	 loss=2.449921	 lr=0.010000
Epoch[007] Batch [0799]/[3760]	Speed: 68.581143 samples/sec	 accuracy=43.306641	 loss=2.453707	 lr=0.010000
Epoch[007] Batch [0849]/[3760]	Speed: 66.058158 samples/sec	 accuracy=43.255515	 loss=2.455324	 lr=0.010000
Epoch[007] Batch [0899]/[3760]	Speed: 70.375908 samples/sec	 accuracy=43.239583	 loss=2.454266	 lr=0.010000
Epoch[007] Batch [0949]/[3760]	Speed: 67.847561 samples/sec	 accuracy=43.292763	 loss=2.452366	 lr=0.010000
Epoch[007] Batch [0999]/[3760]	Speed: 67.073928 samples/sec	 accuracy=43.326563	 loss=2.451364	 lr=0.010000
Epoch[007] Batch [1049]/[3760]	Speed: 69.693754 samples/sec	 accuracy=43.218750	 loss=2.453783	 lr=0.010000
Epoch[007] Batch [1099]/[3760]	Speed: 67.042165 samples/sec	 accuracy=43.232955	 loss=2.454660	 lr=0.010000
Epoch[007] Batch [1149]/[3760]	Speed: 70.505790 samples/sec	 accuracy=43.235054	 loss=2.453153	 lr=0.010000
Epoch[007] Batch [1199]/[3760]	Speed: 66.981976 samples/sec	 accuracy=43.276042	 loss=2.450897	 lr=0.010000
Epoch[007] Batch [1249]/[3760]	Speed: 69.819155 samples/sec	 accuracy=43.318750	 loss=2.449238	 lr=0.010000
Epoch[007] Batch [1299]/[3760]	Speed: 65.679978 samples/sec	 accuracy=43.287260	 loss=2.448639	 lr=0.010000
Epoch[007] Batch [1349]/[3760]	Speed: 71.088447 samples/sec	 accuracy=43.281250	 loss=2.447470	 lr=0.010000
Epoch[007] Batch [1399]/[3760]	Speed: 69.575406 samples/sec	 accuracy=43.251116	 loss=2.449821	 lr=0.010000
Epoch[007] Batch [1449]/[3760]	Speed: 65.954561 samples/sec	 accuracy=43.276940	 loss=2.448878	 lr=0.010000
Epoch[007] Batch [1499]/[3760]	Speed: 69.449042 samples/sec	 accuracy=43.293750	 loss=2.447936	 lr=0.010000
Epoch[007] Batch [1549]/[3760]	Speed: 69.279756 samples/sec	 accuracy=43.270161	 loss=2.448520	 lr=0.010000
Epoch[007] Batch [1599]/[3760]	Speed: 67.993625 samples/sec	 accuracy=43.271484	 loss=2.448560	 lr=0.010000
Epoch[007] Batch [1649]/[3760]	Speed: 67.672294 samples/sec	 accuracy=43.310606	 loss=2.448000	 lr=0.010000
Epoch[007] Batch [1699]/[3760]	Speed: 68.161441 samples/sec	 accuracy=43.316176	 loss=2.447993	 lr=0.010000
Epoch[007] Batch [1749]/[3760]	Speed: 69.654589 samples/sec	 accuracy=43.328571	 loss=2.448959	 lr=0.010000
Epoch[007] Batch [1799]/[3760]	Speed: 67.804111 samples/sec	 accuracy=43.301215	 loss=2.449126	 lr=0.010000
Epoch[007] Batch [1849]/[3760]	Speed: 67.941482 samples/sec	 accuracy=43.283784	 loss=2.450353	 lr=0.010000
Epoch[007] Batch [1899]/[3760]	Speed: 69.147420 samples/sec	 accuracy=43.299342	 loss=2.449775	 lr=0.010000
Epoch[007] Batch [1949]/[3760]	Speed: 66.548241 samples/sec	 accuracy=43.298878	 loss=2.450438	 lr=0.010000
Epoch[007] Batch [1999]/[3760]	Speed: 69.091555 samples/sec	 accuracy=43.274219	 loss=2.451025	 lr=0.010000
Epoch[007] Batch [2049]/[3760]	Speed: 69.007241 samples/sec	 accuracy=43.276677	 loss=2.452166	 lr=0.010000
Epoch[007] Batch [2099]/[3760]	Speed: 68.313021 samples/sec	 accuracy=43.299107	 loss=2.452245	 lr=0.010000
Epoch[007] Batch [2149]/[3760]	Speed: 68.571796 samples/sec	 accuracy=43.282703	 loss=2.453047	 lr=0.010000
Epoch[007] Batch [2199]/[3760]	Speed: 70.752818 samples/sec	 accuracy=43.285511	 loss=2.453489	 lr=0.010000
Epoch[007] Batch [2249]/[3760]	Speed: 67.634044 samples/sec	 accuracy=43.295833	 loss=2.452860	 lr=0.010000
Epoch[007] Batch [2299]/[3760]	Speed: 65.511811 samples/sec	 accuracy=43.313859	 loss=2.452251	 lr=0.010000
Epoch[007] Batch [2349]/[3760]	Speed: 71.740378 samples/sec	 accuracy=43.303191	 loss=2.453132	 lr=0.010000
Epoch[007] Batch [2399]/[3760]	Speed: 68.250185 samples/sec	 accuracy=43.285156	 loss=2.453431	 lr=0.010000
Epoch[007] Batch [2449]/[3760]	Speed: 66.315081 samples/sec	 accuracy=43.288265	 loss=2.453591	 lr=0.010000
Epoch[007] Batch [2499]/[3760]	Speed: 69.860841 samples/sec	 accuracy=43.263750	 loss=2.454636	 lr=0.010000
Epoch[007] Batch [2549]/[3760]	Speed: 69.668548 samples/sec	 accuracy=43.267770	 loss=2.455153	 lr=0.010000
Epoch[007] Batch [2599]/[3760]	Speed: 66.961467 samples/sec	 accuracy=43.270433	 loss=2.454949	 lr=0.010000
Epoch[007] Batch [2649]/[3760]	Speed: 69.470652 samples/sec	 accuracy=43.288325	 loss=2.455144	 lr=0.010000
Epoch[007] Batch [2699]/[3760]	Speed: 70.792284 samples/sec	 accuracy=43.284144	 loss=2.455175	 lr=0.010000
Epoch[007] Batch [2749]/[3760]	Speed: 67.412162 samples/sec	 accuracy=43.274432	 loss=2.455821	 lr=0.010000
Epoch[007] Batch [2799]/[3760]	Speed: 68.451893 samples/sec	 accuracy=43.265625	 loss=2.455764	 lr=0.010000
Epoch[007] Batch [2849]/[3760]	Speed: 69.783769 samples/sec	 accuracy=43.265351	 loss=2.455630	 lr=0.010000
Epoch[007] Batch [2899]/[3760]	Speed: 68.377760 samples/sec	 accuracy=43.273168	 loss=2.455287	 lr=0.010000
Epoch[007] Batch [2949]/[3760]	Speed: 70.298859 samples/sec	 accuracy=43.272246	 loss=2.454497	 lr=0.010000
Epoch[007] Batch [2999]/[3760]	Speed: 68.309294 samples/sec	 accuracy=43.285937	 loss=2.454545	 lr=0.010000
Epoch[007] Batch [3049]/[3760]	Speed: 69.447709 samples/sec	 accuracy=43.307889	 loss=2.454578	 lr=0.010000
Epoch[007] Batch [3099]/[3760]	Speed: 67.075271 samples/sec	 accuracy=43.315524	 loss=2.454245	 lr=0.010000
Epoch[007] Batch [3149]/[3760]	Speed: 68.823547 samples/sec	 accuracy=43.324405	 loss=2.453888	 lr=0.010000
Epoch[007] Batch [3199]/[3760]	Speed: 70.111725 samples/sec	 accuracy=43.335938	 loss=2.453655	 lr=0.010000
Epoch[007] Batch [3249]/[3760]	Speed: 69.213126 samples/sec	 accuracy=43.338462	 loss=2.453520	 lr=0.010000
Epoch[007] Batch [3299]/[3760]	Speed: 66.195991 samples/sec	 accuracy=43.361269	 loss=2.452060	 lr=0.010000
Epoch[007] Batch [3349]/[3760]	Speed: 69.715349 samples/sec	 accuracy=43.347481	 loss=2.452615	 lr=0.010000
Epoch[007] Batch [3399]/[3760]	Speed: 68.812669 samples/sec	 accuracy=43.328125	 loss=2.452643	 lr=0.010000
Epoch[007] Batch [3449]/[3760]	Speed: 68.974101 samples/sec	 accuracy=43.337862	 loss=2.453033	 lr=0.010000
Epoch[007] Batch [3499]/[3760]	Speed: 69.162639 samples/sec	 accuracy=43.341964	 loss=2.451995	 lr=0.010000
Epoch[007] Batch [3549]/[3760]	Speed: 66.727691 samples/sec	 accuracy=43.355194	 loss=2.451030	 lr=0.010000
Epoch[007] Batch [3599]/[3760]	Speed: 69.686703 samples/sec	 accuracy=43.339410	 loss=2.451809	 lr=0.010000
Epoch[007] Batch [3649]/[3760]	Speed: 68.638617 samples/sec	 accuracy=43.353596	 loss=2.451913	 lr=0.010000
Epoch[007] Batch [3699]/[3760]	Speed: 69.342656 samples/sec	 accuracy=43.341216	 loss=2.452272	 lr=0.010000
Epoch[007] Batch [3749]/[3760]	Speed: 76.269701 samples/sec	 accuracy=43.340417	 loss=2.451739	 lr=0.010000
Batch [0049]/[0303]: acc-top1=46.375000 acc-top5=72.500000
Batch [0099]/[0303]: acc-top1=45.968750 acc-top5=72.500000
Batch [0149]/[0303]: acc-top1=46.479167 acc-top5=72.875000
Batch [0199]/[0303]: acc-top1=46.812500 acc-top5=73.203125
Batch [0249]/[0303]: acc-top1=46.893750 acc-top5=73.300000
Batch [0299]/[0303]: acc-top1=46.546875 acc-top5=73.083333
[Epoch 007] training: accuracy=43.338182	 loss=2.451599
[Epoch 007] speed: 68 samples/sec	time cost: 3837.798260
[Epoch 007] validation: acc-top1=46.544967 acc-top5=73.071370 loss=2.458205
Epoch[008] Batch [0049]/[3759]	Speed: 47.642556 samples/sec	 accuracy=45.906250	 loss=2.332233	 lr=0.010000
Epoch[008] Batch [0099]/[3759]	Speed: 69.993197 samples/sec	 accuracy=45.546875	 loss=2.350630	 lr=0.010000
Epoch[008] Batch [0149]/[3759]	Speed: 66.788415 samples/sec	 accuracy=45.312500	 loss=2.346339	 lr=0.010000
Epoch[008] Batch [0199]/[3759]	Speed: 69.096422 samples/sec	 accuracy=45.117188	 loss=2.349392	 lr=0.010000
Epoch[008] Batch [0249]/[3759]	Speed: 67.213045 samples/sec	 accuracy=45.225000	 loss=2.358194	 lr=0.010000
Epoch[008] Batch [0299]/[3759]	Speed: 69.858634 samples/sec	 accuracy=45.286458	 loss=2.358756	 lr=0.010000
Epoch[008] Batch [0349]/[3759]	Speed: 69.099962 samples/sec	 accuracy=45.133929	 loss=2.370199	 lr=0.010000
Epoch[008] Batch [0399]/[3759]	Speed: 67.351608 samples/sec	 accuracy=44.867188	 loss=2.380544	 lr=0.010000
Epoch[008] Batch [0449]/[3759]	Speed: 68.714184 samples/sec	 accuracy=44.802083	 loss=2.378916	 lr=0.010000
Epoch[008] Batch [0499]/[3759]	Speed: 69.629824 samples/sec	 accuracy=44.721875	 loss=2.384482	 lr=0.010000
Epoch[008] Batch [0549]/[3759]	Speed: 69.337435 samples/sec	 accuracy=44.676136	 loss=2.384537	 lr=0.010000
Epoch[008] Batch [0599]/[3759]	Speed: 68.544475 samples/sec	 accuracy=44.643229	 loss=2.382906	 lr=0.010000
Epoch[008] Batch [0649]/[3759]	Speed: 69.221471 samples/sec	 accuracy=44.711538	 loss=2.384693	 lr=0.010000
Epoch[008] Batch [0699]/[3759]	Speed: 68.596248 samples/sec	 accuracy=44.678571	 loss=2.390449	 lr=0.010000
Epoch[008] Batch [0749]/[3759]	Speed: 67.359698 samples/sec	 accuracy=44.750000	 loss=2.387493	 lr=0.010000
Epoch[008] Batch [0799]/[3759]	Speed: 68.680895 samples/sec	 accuracy=44.845703	 loss=2.383747	 lr=0.010000
Epoch[008] Batch [0849]/[3759]	Speed: 69.165272 samples/sec	 accuracy=44.788603	 loss=2.383497	 lr=0.010000
Epoch[008] Batch [0899]/[3759]	Speed: 69.065183 samples/sec	 accuracy=44.772569	 loss=2.386525	 lr=0.010000
Epoch[008] Batch [0949]/[3759]	Speed: 69.348083 samples/sec	 accuracy=44.787829	 loss=2.388167	 lr=0.010000
Epoch[008] Batch [0999]/[3759]	Speed: 67.897210 samples/sec	 accuracy=44.795313	 loss=2.387502	 lr=0.010000
Epoch[008] Batch [1049]/[3759]	Speed: 68.557214 samples/sec	 accuracy=44.812500	 loss=2.386333	 lr=0.010000
Epoch[008] Batch [1099]/[3759]	Speed: 69.490679 samples/sec	 accuracy=44.796875	 loss=2.386635	 lr=0.010000
Epoch[008] Batch [1149]/[3759]	Speed: 67.488930 samples/sec	 accuracy=44.752717	 loss=2.386549	 lr=0.010000
Epoch[008] Batch [1199]/[3759]	Speed: 70.037637 samples/sec	 accuracy=44.654948	 loss=2.389727	 lr=0.010000
Epoch[008] Batch [1249]/[3759]	Speed: 69.582217 samples/sec	 accuracy=44.631250	 loss=2.390173	 lr=0.010000
Epoch[008] Batch [1299]/[3759]	Speed: 68.909115 samples/sec	 accuracy=44.590144	 loss=2.392625	 lr=0.010000
Epoch[008] Batch [1349]/[3759]	Speed: 68.572620 samples/sec	 accuracy=44.655093	 loss=2.392596	 lr=0.010000
Epoch[008] Batch [1399]/[3759]	Speed: 68.636825 samples/sec	 accuracy=44.635045	 loss=2.393576	 lr=0.010000
Epoch[008] Batch [1449]/[3759]	Speed: 67.984421 samples/sec	 accuracy=44.626078	 loss=2.393659	 lr=0.010000
Epoch[008] Batch [1499]/[3759]	Speed: 66.448538 samples/sec	 accuracy=44.654167	 loss=2.391562	 lr=0.010000
Epoch[008] Batch [1549]/[3759]	Speed: 71.614346 samples/sec	 accuracy=44.664315	 loss=2.391386	 lr=0.010000
Epoch[008] Batch [1599]/[3759]	Speed: 68.778472 samples/sec	 accuracy=44.658203	 loss=2.392045	 lr=0.010000
Epoch[008] Batch [1649]/[3759]	Speed: 68.913113 samples/sec	 accuracy=44.639205	 loss=2.391793	 lr=0.010000
Epoch[008] Batch [1699]/[3759]	Speed: 69.341478 samples/sec	 accuracy=44.587316	 loss=2.393492	 lr=0.010000
Epoch[008] Batch [1749]/[3759]	Speed: 68.467642 samples/sec	 accuracy=44.606250	 loss=2.394129	 lr=0.010000
Epoch[008] Batch [1799]/[3759]	Speed: 67.787481 samples/sec	 accuracy=44.570312	 loss=2.394963	 lr=0.010000
Epoch[008] Batch [1849]/[3759]	Speed: 69.330009 samples/sec	 accuracy=44.561655	 loss=2.395555	 lr=0.010000
Epoch[008] Batch [1899]/[3759]	Speed: 69.905780 samples/sec	 accuracy=44.585526	 loss=2.395725	 lr=0.010000
Epoch[008] Batch [1949]/[3759]	Speed: 67.573003 samples/sec	 accuracy=44.587340	 loss=2.395864	 lr=0.010000
Epoch[008] Batch [1999]/[3759]	Speed: 69.488271 samples/sec	 accuracy=44.581250	 loss=2.396055	 lr=0.010000
Epoch[008] Batch [2049]/[3759]	Speed: 66.692070 samples/sec	 accuracy=44.590701	 loss=2.395390	 lr=0.010000
Epoch[008] Batch [2099]/[3759]	Speed: 71.313202 samples/sec	 accuracy=44.617560	 loss=2.395078	 lr=0.010000
Epoch[008] Batch [2149]/[3759]	Speed: 68.549886 samples/sec	 accuracy=44.652616	 loss=2.393427	 lr=0.010000
Epoch[008] Batch [2199]/[3759]	Speed: 66.802054 samples/sec	 accuracy=44.650568	 loss=2.394086	 lr=0.010000
Epoch[008] Batch [2249]/[3759]	Speed: 71.002901 samples/sec	 accuracy=44.649306	 loss=2.394405	 lr=0.010000
Epoch[008] Batch [2299]/[3759]	Speed: 68.874846 samples/sec	 accuracy=44.667120	 loss=2.394233	 lr=0.010000
Epoch[008] Batch [2349]/[3759]	Speed: 67.863515 samples/sec	 accuracy=44.662234	 loss=2.394031	 lr=0.010000
Epoch[008] Batch [2399]/[3759]	Speed: 67.640650 samples/sec	 accuracy=44.679036	 loss=2.394305	 lr=0.010000
Epoch[008] Batch [2449]/[3759]	Speed: 67.989992 samples/sec	 accuracy=44.669643	 loss=2.394217	 lr=0.010000
Epoch[008] Batch [2499]/[3759]	Speed: 69.980071 samples/sec	 accuracy=44.655625	 loss=2.394095	 lr=0.010000
Epoch[008] Batch [2549]/[3759]	Speed: 69.712055 samples/sec	 accuracy=44.657475	 loss=2.393984	 lr=0.010000
Epoch[008] Batch [2599]/[3759]	Speed: 68.774989 samples/sec	 accuracy=44.681490	 loss=2.393088	 lr=0.010000
Epoch[008] Batch [2649]/[3759]	Speed: 69.523635 samples/sec	 accuracy=44.673939	 loss=2.392966	 lr=0.010000
Epoch[008] Batch [2699]/[3759]	Speed: 69.874653 samples/sec	 accuracy=44.682870	 loss=2.392232	 lr=0.010000
Epoch[008] Batch [2749]/[3759]	Speed: 68.177450 samples/sec	 accuracy=44.688636	 loss=2.392109	 lr=0.010000
Epoch[008] Batch [2799]/[3759]	Speed: 68.147853 samples/sec	 accuracy=44.692522	 loss=2.391407	 lr=0.010000
Epoch[008] Batch [2849]/[3759]	Speed: 69.342683 samples/sec	 accuracy=44.666118	 loss=2.392926	 lr=0.010000
Epoch[008] Batch [2899]/[3759]	Speed: 70.681863 samples/sec	 accuracy=44.671875	 loss=2.392752	 lr=0.010000
Epoch[008] Batch [2949]/[3759]	Speed: 66.596146 samples/sec	 accuracy=44.669492	 loss=2.392830	 lr=0.010000
Epoch[008] Batch [2999]/[3759]	Speed: 70.034366 samples/sec	 accuracy=44.684375	 loss=2.392026	 lr=0.010000
Epoch[008] Batch [3049]/[3759]	Speed: 69.393139 samples/sec	 accuracy=44.684426	 loss=2.392746	 lr=0.010000
Epoch[008] Batch [3099]/[3759]	Speed: 69.109251 samples/sec	 accuracy=44.683468	 loss=2.392798	 lr=0.010000
Epoch[008] Batch [3149]/[3759]	Speed: 66.485597 samples/sec	 accuracy=44.685516	 loss=2.391976	 lr=0.010000
Epoch[008] Batch [3199]/[3759]	Speed: 69.224108 samples/sec	 accuracy=44.678711	 loss=2.391806	 lr=0.010000
Epoch[008] Batch [3249]/[3759]	Speed: 69.572463 samples/sec	 accuracy=44.676442	 loss=2.391372	 lr=0.010000
Epoch[008] Batch [3299]/[3759]	Speed: 68.034331 samples/sec	 accuracy=44.696496	 loss=2.390631	 lr=0.010000
Epoch[008] Batch [3349]/[3759]	Speed: 70.492804 samples/sec	 accuracy=44.686567	 loss=2.390728	 lr=0.010000
Epoch[008] Batch [3399]/[3759]	Speed: 68.019477 samples/sec	 accuracy=44.689798	 loss=2.390278	 lr=0.010000
Epoch[008] Batch [3449]/[3759]	Speed: 68.770396 samples/sec	 accuracy=44.685688	 loss=2.389877	 lr=0.010000
Epoch[008] Batch [3499]/[3759]	Speed: 68.180140 samples/sec	 accuracy=44.690179	 loss=2.389928	 lr=0.010000
Epoch[008] Batch [3549]/[3759]	Speed: 69.524061 samples/sec	 accuracy=44.701585	 loss=2.389752	 lr=0.010000
Epoch[008] Batch [3599]/[3759]	Speed: 68.418930 samples/sec	 accuracy=44.665365	 loss=2.390744	 lr=0.010000
Epoch[008] Batch [3649]/[3759]	Speed: 69.311111 samples/sec	 accuracy=44.657106	 loss=2.391201	 lr=0.010000
Epoch[008] Batch [3699]/[3759]	Speed: 68.593468 samples/sec	 accuracy=44.654983	 loss=2.391903	 lr=0.010000
Epoch[008] Batch [3749]/[3759]	Speed: 75.150582 samples/sec	 accuracy=44.661250	 loss=2.392230	 lr=0.010000
Batch [0049]/[0303]: acc-top1=47.343750 acc-top5=73.937500
Batch [0099]/[0303]: acc-top1=46.437500 acc-top5=73.656250
Batch [0149]/[0303]: acc-top1=46.718750 acc-top5=73.958333
Batch [0199]/[0303]: acc-top1=46.796875 acc-top5=74.140625
Batch [0249]/[0303]: acc-top1=46.837500 acc-top5=74.175000
Batch [0299]/[0303]: acc-top1=46.750000 acc-top5=73.822917
[Epoch 008] training: accuracy=44.662809	 loss=2.392111
[Epoch 008] speed: 68 samples/sec	time cost: 3820.568052
[Epoch 008] validation: acc-top1=46.771865 acc-top5=73.788160 loss=2.402559
Epoch[009] Batch [0049]/[3760]	Speed: 49.285047 samples/sec	 accuracy=45.156250	 loss=2.376749	 lr=0.010000
Epoch[009] Batch [0099]/[3760]	Speed: 66.835580 samples/sec	 accuracy=45.390625	 loss=2.369885	 lr=0.010000
Epoch[009] Batch [0149]/[3760]	Speed: 69.094514 samples/sec	 accuracy=45.791667	 loss=2.346775	 lr=0.010000
Epoch[009] Batch [0199]/[3760]	Speed: 67.244151 samples/sec	 accuracy=46.085938	 loss=2.342141	 lr=0.010000
Epoch[009] Batch [0249]/[3760]	Speed: 69.616109 samples/sec	 accuracy=45.937500	 loss=2.348808	 lr=0.010000
Epoch[009] Batch [0299]/[3760]	Speed: 69.080476 samples/sec	 accuracy=45.703125	 loss=2.348241	 lr=0.010000
Epoch[009] Batch [0349]/[3760]	Speed: 68.344033 samples/sec	 accuracy=45.732143	 loss=2.343249	 lr=0.010000
Epoch[009] Batch [0399]/[3760]	Speed: 69.290901 samples/sec	 accuracy=45.894531	 loss=2.342623	 lr=0.010000
Epoch[009] Batch [0449]/[3760]	Speed: 69.146352 samples/sec	 accuracy=45.892361	 loss=2.343259	 lr=0.010000
Epoch[009] Batch [0499]/[3760]	Speed: 69.348815 samples/sec	 accuracy=45.909375	 loss=2.340509	 lr=0.010000
Epoch[009] Batch [0549]/[3760]	Speed: 69.995354 samples/sec	 accuracy=45.832386	 loss=2.341695	 lr=0.010000
Epoch[009] Batch [0599]/[3760]	Speed: 66.924254 samples/sec	 accuracy=45.880208	 loss=2.338562	 lr=0.010000
Epoch[009] Batch [0649]/[3760]	Speed: 69.834676 samples/sec	 accuracy=45.865385	 loss=2.337324	 lr=0.010000
Epoch[009] Batch [0699]/[3760]	Speed: 68.035272 samples/sec	 accuracy=45.868304	 loss=2.335387	 lr=0.010000
Epoch[009] Batch [0749]/[3760]	Speed: 70.541939 samples/sec	 accuracy=45.927083	 loss=2.332590	 lr=0.010000
Epoch[009] Batch [0799]/[3760]	Speed: 68.598145 samples/sec	 accuracy=45.925781	 loss=2.332010	 lr=0.010000
Epoch[009] Batch [0849]/[3760]	Speed: 70.625954 samples/sec	 accuracy=45.856618	 loss=2.332934	 lr=0.010000
Epoch[009] Batch [0899]/[3760]	Speed: 66.667143 samples/sec	 accuracy=45.880208	 loss=2.333573	 lr=0.010000
Epoch[009] Batch [0949]/[3760]	Speed: 70.243965 samples/sec	 accuracy=45.924342	 loss=2.331832	 lr=0.010000
Epoch[009] Batch [0999]/[3760]	Speed: 68.199798 samples/sec	 accuracy=45.954688	 loss=2.328544	 lr=0.010000
Epoch[009] Batch [1049]/[3760]	Speed: 69.352933 samples/sec	 accuracy=45.916667	 loss=2.327155	 lr=0.010000
Epoch[009] Batch [1099]/[3760]	Speed: 69.189756 samples/sec	 accuracy=45.964489	 loss=2.327396	 lr=0.010000
Epoch[009] Batch [1149]/[3760]	Speed: 69.346594 samples/sec	 accuracy=45.906250	 loss=2.329267	 lr=0.010000
Epoch[009] Batch [1199]/[3760]	Speed: 69.825743 samples/sec	 accuracy=45.861979	 loss=2.329814	 lr=0.010000
Epoch[009] Batch [1249]/[3760]	Speed: 67.271335 samples/sec	 accuracy=45.856250	 loss=2.330607	 lr=0.010000
Epoch[009] Batch [1299]/[3760]	Speed: 70.964927 samples/sec	 accuracy=45.846154	 loss=2.330028	 lr=0.010000
Epoch[009] Batch [1349]/[3760]	Speed: 68.978112 samples/sec	 accuracy=45.798611	 loss=2.330921	 lr=0.010000
Epoch[009] Batch [1399]/[3760]	Speed: 70.257127 samples/sec	 accuracy=45.809152	 loss=2.332596	 lr=0.010000
Epoch[009] Batch [1449]/[3760]	Speed: 68.033892 samples/sec	 accuracy=45.786638	 loss=2.333768	 lr=0.010000
Epoch[009] Batch [1499]/[3760]	Speed: 69.198197 samples/sec	 accuracy=45.738542	 loss=2.336338	 lr=0.010000
Epoch[009] Batch [1549]/[3760]	Speed: 67.246482 samples/sec	 accuracy=45.737903	 loss=2.336610	 lr=0.010000
Epoch[009] Batch [1599]/[3760]	Speed: 68.711834 samples/sec	 accuracy=45.763672	 loss=2.336313	 lr=0.010000
Epoch[009] Batch [1649]/[3760]	Speed: 68.143389 samples/sec	 accuracy=45.744318	 loss=2.336667	 lr=0.010000
Epoch[009] Batch [1699]/[3760]	Speed: 67.686614 samples/sec	 accuracy=45.713235	 loss=2.338692	 lr=0.010000
Epoch[009] Batch [1749]/[3760]	Speed: 69.909719 samples/sec	 accuracy=45.679464	 loss=2.339970	 lr=0.010000
Epoch[009] Batch [1799]/[3760]	Speed: 69.543819 samples/sec	 accuracy=45.674479	 loss=2.340946	 lr=0.010000
Epoch[009] Batch [1849]/[3760]	Speed: 68.494455 samples/sec	 accuracy=45.706081	 loss=2.340480	 lr=0.010000
Epoch[009] Batch [1899]/[3760]	Speed: 70.974905 samples/sec	 accuracy=45.697368	 loss=2.339786	 lr=0.010000
Epoch[009] Batch [1949]/[3760]	Speed: 68.842273 samples/sec	 accuracy=45.704327	 loss=2.341419	 lr=0.010000
Epoch[009] Batch [1999]/[3760]	Speed: 69.295772 samples/sec	 accuracy=45.696875	 loss=2.341429	 lr=0.010000
Epoch[009] Batch [2049]/[3760]	Speed: 66.888723 samples/sec	 accuracy=45.680640	 loss=2.341293	 lr=0.010000
Epoch[009] Batch [2099]/[3760]	Speed: 70.024742 samples/sec	 accuracy=45.687500	 loss=2.341204	 lr=0.010000
Epoch[009] Batch [2149]/[3760]	Speed: 64.714739 samples/sec	 accuracy=45.651890	 loss=2.342002	 lr=0.010000
Epoch[009] Batch [2199]/[3760]	Speed: 72.501940 samples/sec	 accuracy=45.608665	 loss=2.343901	 lr=0.010000
Epoch[009] Batch [2249]/[3760]	Speed: 67.794919 samples/sec	 accuracy=45.643750	 loss=2.342425	 lr=0.010000
Epoch[009] Batch [2299]/[3760]	Speed: 68.198138 samples/sec	 accuracy=45.650136	 loss=2.342414	 lr=0.010000
Epoch[009] Batch [2349]/[3760]	Speed: 69.682045 samples/sec	 accuracy=45.674867	 loss=2.342128	 lr=0.010000
Epoch[009] Batch [2399]/[3760]	Speed: 70.297843 samples/sec	 accuracy=45.669271	 loss=2.341788	 lr=0.010000
Epoch[009] Batch [2449]/[3760]	Speed: 68.615650 samples/sec	 accuracy=45.685587	 loss=2.341937	 lr=0.010000
Epoch[009] Batch [2499]/[3760]	Speed: 67.648453 samples/sec	 accuracy=45.662500	 loss=2.343229	 lr=0.010000
Epoch[009] Batch [2549]/[3760]	Speed: 70.654746 samples/sec	 accuracy=45.692402	 loss=2.342765	 lr=0.010000
Epoch[009] Batch [2599]/[3760]	Speed: 68.329695 samples/sec	 accuracy=45.694712	 loss=2.342325	 lr=0.010000
Epoch[009] Batch [2649]/[3760]	Speed: 67.398896 samples/sec	 accuracy=45.694575	 loss=2.342685	 lr=0.010000
Epoch[009] Batch [2699]/[3760]	Speed: 69.418018 samples/sec	 accuracy=45.718171	 loss=2.342002	 lr=0.010000
Epoch[009] Batch [2749]/[3760]	Speed: 68.404112 samples/sec	 accuracy=45.715341	 loss=2.341027	 lr=0.010000
Epoch[009] Batch [2799]/[3760]	Speed: 71.029132 samples/sec	 accuracy=45.689732	 loss=2.342337	 lr=0.010000
Epoch[009] Batch [2849]/[3760]	Speed: 68.361602 samples/sec	 accuracy=45.698465	 loss=2.342166	 lr=0.010000
Epoch[009] Batch [2899]/[3760]	Speed: 69.371152 samples/sec	 accuracy=45.721983	 loss=2.341358	 lr=0.010000
Epoch[009] Batch [2949]/[3760]	Speed: 67.766511 samples/sec	 accuracy=45.765890	 loss=2.340494	 lr=0.010000
Epoch[009] Batch [2999]/[3760]	Speed: 69.497836 samples/sec	 accuracy=45.760937	 loss=2.340219	 lr=0.010000
Epoch[009] Batch [3049]/[3760]	Speed: 70.505359 samples/sec	 accuracy=45.752561	 loss=2.340991	 lr=0.010000
Epoch[009] Batch [3099]/[3760]	Speed: 70.493706 samples/sec	 accuracy=45.759577	 loss=2.340538	 lr=0.010000
Epoch[009] Batch [3149]/[3760]	Speed: 66.995839 samples/sec	 accuracy=45.759921	 loss=2.340286	 lr=0.010000
Epoch[009] Batch [3199]/[3760]	Speed: 68.659116 samples/sec	 accuracy=45.791504	 loss=2.339470	 lr=0.010000
Epoch[009] Batch [3249]/[3760]	Speed: 66.570385 samples/sec	 accuracy=45.773558	 loss=2.340348	 lr=0.010000
Epoch[009] Batch [3299]/[3760]	Speed: 70.806027 samples/sec	 accuracy=45.763258	 loss=2.339558	 lr=0.010000
Epoch[009] Batch [3349]/[3760]	Speed: 70.046510 samples/sec	 accuracy=45.768190	 loss=2.339482	 lr=0.010000
Epoch[009] Batch [3399]/[3760]	Speed: 67.968259 samples/sec	 accuracy=45.768382	 loss=2.339642	 lr=0.010000
Epoch[009] Batch [3449]/[3760]	Speed: 69.590134 samples/sec	 accuracy=45.772645	 loss=2.338568	 lr=0.010000
Epoch[009] Batch [3499]/[3760]	Speed: 69.382921 samples/sec	 accuracy=45.757143	 loss=2.339201	 lr=0.010000
Epoch[009] Batch [3549]/[3760]	Speed: 66.252188 samples/sec	 accuracy=45.769806	 loss=2.339299	 lr=0.010000
Epoch[009] Batch [3599]/[3760]	Speed: 67.604069 samples/sec	 accuracy=45.740017	 loss=2.340743	 lr=0.010000
Epoch[009] Batch [3649]/[3760]	Speed: 68.826137 samples/sec	 accuracy=45.732021	 loss=2.341099	 lr=0.010000
Epoch[009] Batch [3699]/[3760]	Speed: 72.392363 samples/sec	 accuracy=45.754223	 loss=2.340362	 lr=0.010000
Epoch[009] Batch [3749]/[3760]	Speed: 76.600518 samples/sec	 accuracy=45.750833	 loss=2.339984	 lr=0.010000
Batch [0049]/[0303]: acc-top1=48.312500 acc-top5=74.562500
Batch [0099]/[0303]: acc-top1=47.937500 acc-top5=74.281250
Batch [0149]/[0303]: acc-top1=48.645833 acc-top5=74.885417
Batch [0199]/[0303]: acc-top1=49.046875 acc-top5=74.929688
Batch [0249]/[0303]: acc-top1=49.068750 acc-top5=75.106250
Batch [0299]/[0303]: acc-top1=48.708333 acc-top5=74.760417
[Epoch 009] training: accuracy=45.751745	 loss=2.339994
[Epoch 009] speed: 68 samples/sec	time cost: 3813.492530
[Epoch 009] validation: acc-top1=48.741749 acc-top5=74.726691 loss=2.281452
Epoch[010] Batch [0049]/[3760]	Speed: 47.997144 samples/sec	 accuracy=46.375000	 loss=2.321493	 lr=0.010000
Epoch[010] Batch [0099]/[3760]	Speed: 65.770643 samples/sec	 accuracy=46.359375	 loss=2.324095	 lr=0.010000
Epoch[010] Batch [0149]/[3760]	Speed: 70.201101 samples/sec	 accuracy=46.645833	 loss=2.306073	 lr=0.010000
Epoch[010] Batch [0199]/[3760]	Speed: 68.570926 samples/sec	 accuracy=46.335938	 loss=2.316041	 lr=0.010000
Epoch[010] Batch [0249]/[3760]	Speed: 69.350277 samples/sec	 accuracy=46.243750	 loss=2.308294	 lr=0.010000
Epoch[010] Batch [0299]/[3760]	Speed: 68.940260 samples/sec	 accuracy=46.442708	 loss=2.299264	 lr=0.010000
Epoch[010] Batch [0349]/[3760]	Speed: 69.525515 samples/sec	 accuracy=46.513393	 loss=2.298165	 lr=0.010000
Epoch[010] Batch [0399]/[3760]	Speed: 70.221784 samples/sec	 accuracy=46.597656	 loss=2.295629	 lr=0.010000
Epoch[010] Batch [0449]/[3760]	Speed: 68.279685 samples/sec	 accuracy=46.552083	 loss=2.297630	 lr=0.010000
Epoch[010] Batch [0499]/[3760]	Speed: 70.005757 samples/sec	 accuracy=46.556250	 loss=2.295132	 lr=0.010000
Epoch[010] Batch [0549]/[3760]	Speed: 67.886401 samples/sec	 accuracy=46.451705	 loss=2.299360	 lr=0.010000
Epoch[010] Batch [0599]/[3760]	Speed: 69.236625 samples/sec	 accuracy=46.416667	 loss=2.298670	 lr=0.010000
Epoch[010] Batch [0649]/[3760]	Speed: 70.103873 samples/sec	 accuracy=46.319712	 loss=2.303365	 lr=0.010000
Epoch[010] Batch [0699]/[3760]	Speed: 67.707648 samples/sec	 accuracy=46.290179	 loss=2.305244	 lr=0.010000
Epoch[010] Batch [0749]/[3760]	Speed: 70.466400 samples/sec	 accuracy=46.327083	 loss=2.301960	 lr=0.010000
Epoch[010] Batch [0799]/[3760]	Speed: 69.401604 samples/sec	 accuracy=46.425781	 loss=2.299454	 lr=0.010000
Epoch[010] Batch [0849]/[3760]	Speed: 67.244268 samples/sec	 accuracy=46.378676	 loss=2.300521	 lr=0.010000
Epoch[010] Batch [0899]/[3760]	Speed: 71.113560 samples/sec	 accuracy=46.435764	 loss=2.301986	 lr=0.010000
Epoch[010] Batch [0949]/[3760]	Speed: 69.196344 samples/sec	 accuracy=46.457237	 loss=2.302398	 lr=0.010000
Epoch[010] Batch [0999]/[3760]	Speed: 67.194370 samples/sec	 accuracy=46.432812	 loss=2.306184	 lr=0.010000
Epoch[010] Batch [1049]/[3760]	Speed: 70.394296 samples/sec	 accuracy=46.389881	 loss=2.308361	 lr=0.010000
Epoch[010] Batch [1099]/[3760]	Speed: 69.858207 samples/sec	 accuracy=46.444602	 loss=2.304402	 lr=0.010000
Epoch[010] Batch [1149]/[3760]	Speed: 68.802193 samples/sec	 accuracy=46.429348	 loss=2.302595	 lr=0.010000
Epoch[010] Batch [1199]/[3760]	Speed: 70.446839 samples/sec	 accuracy=46.458333	 loss=2.301156	 lr=0.010000
Epoch[010] Batch [1249]/[3760]	Speed: 66.671753 samples/sec	 accuracy=46.526250	 loss=2.298565	 lr=0.010000
Epoch[010] Batch [1299]/[3760]	Speed: 69.773301 samples/sec	 accuracy=46.581731	 loss=2.296062	 lr=0.010000
Epoch[010] Batch [1349]/[3760]	Speed: 70.254592 samples/sec	 accuracy=46.587963	 loss=2.297397	 lr=0.010000
Epoch[010] Batch [1399]/[3760]	Speed: 66.964183 samples/sec	 accuracy=46.580357	 loss=2.297515	 lr=0.010000
Epoch[010] Batch [1449]/[3760]	Speed: 70.423713 samples/sec	 accuracy=46.567888	 loss=2.298524	 lr=0.010000
Epoch[010] Batch [1499]/[3760]	Speed: 70.114907 samples/sec	 accuracy=46.529167	 loss=2.300978	 lr=0.010000
Epoch[010] Batch [1549]/[3760]	Speed: 68.880136 samples/sec	 accuracy=46.510081	 loss=2.301088	 lr=0.010000
Epoch[010] Batch [1599]/[3760]	Speed: 69.037884 samples/sec	 accuracy=46.493164	 loss=2.300991	 lr=0.010000
Epoch[010] Batch [1649]/[3760]	Speed: 66.195293 samples/sec	 accuracy=46.504735	 loss=2.298927	 lr=0.010000
Epoch[010] Batch [1699]/[3760]	Speed: 70.692108 samples/sec	 accuracy=46.477941	 loss=2.299822	 lr=0.010000
Epoch[010] Batch [1749]/[3760]	Speed: 68.154374 samples/sec	 accuracy=46.469643	 loss=2.300091	 lr=0.010000
Epoch[010] Batch [1799]/[3760]	Speed: 69.766182 samples/sec	 accuracy=46.433160	 loss=2.301417	 lr=0.010000
Epoch[010] Batch [1849]/[3760]	Speed: 70.057779 samples/sec	 accuracy=46.401182	 loss=2.300927	 lr=0.010000
Epoch[010] Batch [1899]/[3760]	Speed: 69.308484 samples/sec	 accuracy=46.425987	 loss=2.300863	 lr=0.010000
Epoch[010] Batch [1949]/[3760]	Speed: 69.959754 samples/sec	 accuracy=46.435096	 loss=2.300528	 lr=0.010000
Epoch[010] Batch [1999]/[3760]	Speed: 67.655190 samples/sec	 accuracy=46.435938	 loss=2.299544	 lr=0.010000
Epoch[010] Batch [2049]/[3760]	Speed: 69.552874 samples/sec	 accuracy=46.408537	 loss=2.300209	 lr=0.010000
Epoch[010] Batch [2099]/[3760]	Speed: 70.234212 samples/sec	 accuracy=46.404762	 loss=2.300440	 lr=0.010000
Epoch[010] Batch [2149]/[3760]	Speed: 68.154042 samples/sec	 accuracy=46.446221	 loss=2.299671	 lr=0.010000
Epoch[010] Batch [2199]/[3760]	Speed: 69.692553 samples/sec	 accuracy=46.442472	 loss=2.298857	 lr=0.010000
Epoch[010] Batch [2249]/[3760]	Speed: 69.924938 samples/sec	 accuracy=46.446528	 loss=2.298984	 lr=0.010000
Epoch[010] Batch [2299]/[3760]	Speed: 68.825308 samples/sec	 accuracy=46.463995	 loss=2.298518	 lr=0.010000
Epoch[010] Batch [2349]/[3760]	Speed: 68.204232 samples/sec	 accuracy=46.494016	 loss=2.297350	 lr=0.010000
Epoch[010] Batch [2399]/[3760]	Speed: 68.797772 samples/sec	 accuracy=46.500000	 loss=2.295969	 lr=0.010000
Epoch[010] Batch [2449]/[3760]	Speed: 65.845895 samples/sec	 accuracy=46.500000	 loss=2.295433	 lr=0.010000
Epoch[010] Batch [2499]/[3760]	Speed: 71.798421 samples/sec	 accuracy=46.514375	 loss=2.295852	 lr=0.010000
Epoch[010] Batch [2549]/[3760]	Speed: 69.073915 samples/sec	 accuracy=46.531250	 loss=2.295852	 lr=0.010000
Epoch[010] Batch [2599]/[3760]	Speed: 69.592600 samples/sec	 accuracy=46.502404	 loss=2.296939	 lr=0.010000
Epoch[010] Batch [2649]/[3760]	Speed: 69.453523 samples/sec	 accuracy=46.532429	 loss=2.295179	 lr=0.010000
Epoch[010] Batch [2699]/[3760]	Speed: 69.614008 samples/sec	 accuracy=46.517940	 loss=2.296200	 lr=0.010000
Epoch[010] Batch [2749]/[3760]	Speed: 69.736285 samples/sec	 accuracy=46.517045	 loss=2.296133	 lr=0.010000
Epoch[010] Batch [2799]/[3760]	Speed: 66.832495 samples/sec	 accuracy=46.508929	 loss=2.296730	 lr=0.010000
Epoch[010] Batch [2849]/[3760]	Speed: 71.063986 samples/sec	 accuracy=46.504934	 loss=2.297429	 lr=0.010000
Epoch[010] Batch [2899]/[3760]	Speed: 69.625242 samples/sec	 accuracy=46.495151	 loss=2.297792	 lr=0.010000
Epoch[010] Batch [2949]/[3760]	Speed: 66.676493 samples/sec	 accuracy=46.490466	 loss=2.297431	 lr=0.010000
Epoch[010] Batch [2999]/[3760]	Speed: 71.349964 samples/sec	 accuracy=46.497917	 loss=2.297550	 lr=0.010000
Epoch[010] Batch [3049]/[3760]	Speed: 69.546580 samples/sec	 accuracy=46.503586	 loss=2.297878	 lr=0.010000
Epoch[010] Batch [3099]/[3760]	Speed: 68.075709 samples/sec	 accuracy=46.487903	 loss=2.298745	 lr=0.010000
Epoch[010] Batch [3149]/[3760]	Speed: 69.563758 samples/sec	 accuracy=46.499008	 loss=2.298890	 lr=0.010000
Epoch[010] Batch [3199]/[3760]	Speed: 70.046373 samples/sec	 accuracy=46.526367	 loss=2.298320	 lr=0.010000
Epoch[010] Batch [3249]/[3760]	Speed: 68.623453 samples/sec	 accuracy=46.515865	 loss=2.298354	 lr=0.010000
Epoch[010] Batch [3299]/[3760]	Speed: 68.478992 samples/sec	 accuracy=46.509943	 loss=2.297819	 lr=0.010000
Epoch[010] Batch [3349]/[3760]	Speed: 70.493989 samples/sec	 accuracy=46.500000	 loss=2.297827	 lr=0.010000
Epoch[010] Batch [3399]/[3760]	Speed: 68.426881 samples/sec	 accuracy=46.494485	 loss=2.297911	 lr=0.010000
Epoch[010] Batch [3449]/[3760]	Speed: 69.792915 samples/sec	 accuracy=46.521286	 loss=2.297100	 lr=0.010000
Epoch[010] Batch [3499]/[3760]	Speed: 69.655797 samples/sec	 accuracy=46.507589	 loss=2.297375	 lr=0.010000
Epoch[010] Batch [3549]/[3760]	Speed: 70.161991 samples/sec	 accuracy=46.499120	 loss=2.298094	 lr=0.010000
Epoch[010] Batch [3599]/[3760]	Speed: 68.886782 samples/sec	 accuracy=46.513021	 loss=2.297277	 lr=0.010000
Epoch[010] Batch [3649]/[3760]	Speed: 68.462173 samples/sec	 accuracy=46.525685	 loss=2.296798	 lr=0.010000
Epoch[010] Batch [3699]/[3760]	Speed: 70.379342 samples/sec	 accuracy=46.521115	 loss=2.297138	 lr=0.010000
Epoch[010] Batch [3749]/[3760]	Speed: 76.776468 samples/sec	 accuracy=46.517917	 loss=2.297260	 lr=0.010000
Batch [0049]/[0303]: acc-top1=49.250000 acc-top5=74.125000
Batch [0099]/[0303]: acc-top1=48.296875 acc-top5=73.828125
Batch [0149]/[0303]: acc-top1=48.916667 acc-top5=74.468750
Batch [0199]/[0303]: acc-top1=48.953125 acc-top5=74.625000
Batch [0249]/[0303]: acc-top1=49.006250 acc-top5=74.843750
Batch [0299]/[0303]: acc-top1=48.791667 acc-top5=74.447917
[Epoch 010] training: accuracy=46.510971	 loss=2.297359
[Epoch 010] speed: 68 samples/sec	time cost: 3800.887541
[Epoch 010] validation: acc-top1=48.788160 acc-top5=74.406972 loss=2.329944
Epoch[011] Batch [0049]/[3759]	Speed: 47.441159 samples/sec	 accuracy=47.062500	 loss=2.235708	 lr=0.010000
Epoch[011] Batch [0099]/[3759]	Speed: 67.967479 samples/sec	 accuracy=47.156250	 loss=2.228340	 lr=0.010000
Epoch[011] Batch [0149]/[3759]	Speed: 70.089433 samples/sec	 accuracy=47.468750	 loss=2.232458	 lr=0.010000
Epoch[011] Batch [0199]/[3759]	Speed: 69.390331 samples/sec	 accuracy=47.617188	 loss=2.219249	 lr=0.010000
Epoch[011] Batch [0249]/[3759]	Speed: 68.143168 samples/sec	 accuracy=47.475000	 loss=2.220307	 lr=0.010000
Epoch[011] Batch [0299]/[3759]	Speed: 70.769252 samples/sec	 accuracy=47.432292	 loss=2.227436	 lr=0.010000
Epoch[011] Batch [0349]/[3759]	Speed: 68.176349 samples/sec	 accuracy=47.437500	 loss=2.227781	 lr=0.010000
Epoch[011] Batch [0399]/[3759]	Speed: 70.992621 samples/sec	 accuracy=47.734375	 loss=2.222555	 lr=0.010000
Epoch[011] Batch [0449]/[3759]	Speed: 66.769756 samples/sec	 accuracy=47.597222	 loss=2.225588	 lr=0.010000
Epoch[011] Batch [0499]/[3759]	Speed: 70.806894 samples/sec	 accuracy=47.481250	 loss=2.230479	 lr=0.010000
Epoch[011] Batch [0549]/[3759]	Speed: 69.695552 samples/sec	 accuracy=47.602273	 loss=2.226995	 lr=0.010000
Epoch[011] Batch [0599]/[3759]	Speed: 67.437866 samples/sec	 accuracy=47.604167	 loss=2.226921	 lr=0.010000
Epoch[011] Batch [0649]/[3759]	Speed: 69.566374 samples/sec	 accuracy=47.564904	 loss=2.229422	 lr=0.010000
Epoch[011] Batch [0699]/[3759]	Speed: 70.886326 samples/sec	 accuracy=47.645089	 loss=2.228285	 lr=0.010000
Epoch[011] Batch [0749]/[3759]	Speed: 68.748994 samples/sec	 accuracy=47.577083	 loss=2.230477	 lr=0.010000
Epoch[011] Batch [0799]/[3759]	Speed: 67.477098 samples/sec	 accuracy=47.546875	 loss=2.232987	 lr=0.010000
Epoch[011] Batch [0849]/[3759]	Speed: 70.850862 samples/sec	 accuracy=47.536765	 loss=2.233162	 lr=0.010000
Epoch[011] Batch [0899]/[3759]	Speed: 68.177731 samples/sec	 accuracy=47.486111	 loss=2.236629	 lr=0.010000
Epoch[011] Batch [0949]/[3759]	Speed: 70.305219 samples/sec	 accuracy=47.381579	 loss=2.241536	 lr=0.010000
Epoch[011] Batch [0999]/[3759]	Speed: 69.493637 samples/sec	 accuracy=47.328125	 loss=2.242074	 lr=0.010000
Epoch[011] Batch [1049]/[3759]	Speed: 69.756200 samples/sec	 accuracy=47.340774	 loss=2.243372	 lr=0.010000
Epoch[011] Batch [1099]/[3759]	Speed: 67.832642 samples/sec	 accuracy=47.340909	 loss=2.242486	 lr=0.010000
Epoch[011] Batch [1149]/[3759]	Speed: 70.258792 samples/sec	 accuracy=47.336957	 loss=2.242728	 lr=0.010000
Epoch[011] Batch [1199]/[3759]	Speed: 69.102361 samples/sec	 accuracy=47.346354	 loss=2.244679	 lr=0.010000
Epoch[011] Batch [1249]/[3759]	Speed: 71.317705 samples/sec	 accuracy=47.363750	 loss=2.244520	 lr=0.010000
Epoch[011] Batch [1299]/[3759]	Speed: 69.308895 samples/sec	 accuracy=47.338942	 loss=2.246185	 lr=0.010000
Epoch[011] Batch [1349]/[3759]	Speed: 69.197821 samples/sec	 accuracy=47.314815	 loss=2.247565	 lr=0.010000
Epoch[011] Batch [1399]/[3759]	Speed: 68.680720 samples/sec	 accuracy=47.290179	 loss=2.250221	 lr=0.010000
Epoch[011] Batch [1449]/[3759]	Speed: 70.148092 samples/sec	 accuracy=47.274784	 loss=2.252399	 lr=0.010000
Epoch[011] Batch [1499]/[3759]	Speed: 65.154124 samples/sec	 accuracy=47.275000	 loss=2.252945	 lr=0.010000
Epoch[011] Batch [1549]/[3759]	Speed: 73.035873 samples/sec	 accuracy=47.281250	 loss=2.251890	 lr=0.010000
Epoch[011] Batch [1599]/[3759]	Speed: 70.498086 samples/sec	 accuracy=47.263672	 loss=2.253440	 lr=0.010000
Epoch[011] Batch [1649]/[3759]	Speed: 67.949829 samples/sec	 accuracy=47.287879	 loss=2.252180	 lr=0.010000
Epoch[011] Batch [1699]/[3759]	Speed: 69.092672 samples/sec	 accuracy=47.303309	 loss=2.252982	 lr=0.010000
Epoch[011] Batch [1749]/[3759]	Speed: 69.172097 samples/sec	 accuracy=47.301786	 loss=2.253132	 lr=0.010000
Epoch[011] Batch [1799]/[3759]	Speed: 68.730006 samples/sec	 accuracy=47.333333	 loss=2.253028	 lr=0.010000
Epoch[011] Batch [1849]/[3759]	Speed: 70.758820 samples/sec	 accuracy=47.373311	 loss=2.252106	 lr=0.010000
Epoch[011] Batch [1899]/[3759]	Speed: 69.048796 samples/sec	 accuracy=47.376645	 loss=2.253021	 lr=0.010000
Epoch[011] Batch [1949]/[3759]	Speed: 69.887621 samples/sec	 accuracy=47.387019	 loss=2.252880	 lr=0.010000
Epoch[011] Batch [1999]/[3759]	Speed: 67.666427 samples/sec	 accuracy=47.374219	 loss=2.252943	 lr=0.010000
Epoch[011] Batch [2049]/[3759]	Speed: 68.674900 samples/sec	 accuracy=47.400915	 loss=2.252633	 lr=0.010000
Epoch[011] Batch [2099]/[3759]	Speed: 70.116997 samples/sec	 accuracy=47.404762	 loss=2.252601	 lr=0.010000
Epoch[011] Batch [2149]/[3759]	Speed: 67.487359 samples/sec	 accuracy=47.353924	 loss=2.253787	 lr=0.010000
Epoch[011] Batch [2199]/[3759]	Speed: 71.201661 samples/sec	 accuracy=47.345881	 loss=2.253862	 lr=0.010000
Epoch[011] Batch [2249]/[3759]	Speed: 69.811471 samples/sec	 accuracy=47.360417	 loss=2.253157	 lr=0.010000
Epoch[011] Batch [2299]/[3759]	Speed: 69.509883 samples/sec	 accuracy=47.347826	 loss=2.253775	 lr=0.010000
Epoch[011] Batch [2349]/[3759]	Speed: 69.695070 samples/sec	 accuracy=47.336436	 loss=2.254197	 lr=0.010000
Epoch[011] Batch [2399]/[3759]	Speed: 68.539209 samples/sec	 accuracy=47.345052	 loss=2.253802	 lr=0.010000
Epoch[011] Batch [2449]/[3759]	Speed: 69.363412 samples/sec	 accuracy=47.345663	 loss=2.253556	 lr=0.010000
Epoch[011] Batch [2499]/[3759]	Speed: 68.918001 samples/sec	 accuracy=47.364375	 loss=2.253560	 lr=0.010000
Epoch[011] Batch [2549]/[3759]	Speed: 69.094950 samples/sec	 accuracy=47.362745	 loss=2.254044	 lr=0.010000
Epoch[011] Batch [2599]/[3759]	Speed: 69.725019 samples/sec	 accuracy=47.354567	 loss=2.254363	 lr=0.010000
Epoch[011] Batch [2649]/[3759]	Speed: 69.978993 samples/sec	 accuracy=47.339033	 loss=2.254745	 lr=0.010000
Epoch[011] Batch [2699]/[3759]	Speed: 68.456728 samples/sec	 accuracy=47.326389	 loss=2.254479	 lr=0.010000
Epoch[011] Batch [2749]/[3759]	Speed: 69.441951 samples/sec	 accuracy=47.325568	 loss=2.255143	 lr=0.010000
Epoch[011] Batch [2799]/[3759]	Speed: 68.566801 samples/sec	 accuracy=47.319754	 loss=2.255277	 lr=0.010000
Epoch[011] Batch [2849]/[3759]	Speed: 68.330875 samples/sec	 accuracy=47.315241	 loss=2.255987	 lr=0.010000
Epoch[011] Batch [2899]/[3759]	Speed: 71.120052 samples/sec	 accuracy=47.331358	 loss=2.254308	 lr=0.010000
Epoch[011] Batch [2949]/[3759]	Speed: 69.961971 samples/sec	 accuracy=47.326271	 loss=2.253770	 lr=0.010000
Epoch[011] Batch [2999]/[3759]	Speed: 68.931729 samples/sec	 accuracy=47.341667	 loss=2.253710	 lr=0.010000
Epoch[011] Batch [3049]/[3759]	Speed: 69.028674 samples/sec	 accuracy=47.337602	 loss=2.254690	 lr=0.010000
Epoch[011] Batch [3099]/[3759]	Speed: 70.565072 samples/sec	 accuracy=47.345262	 loss=2.254453	 lr=0.010000
Epoch[011] Batch [3149]/[3759]	Speed: 67.624541 samples/sec	 accuracy=47.334325	 loss=2.255368	 lr=0.010000
Epoch[011] Batch [3199]/[3759]	Speed: 69.981390 samples/sec	 accuracy=47.335938	 loss=2.255773	 lr=0.010000
Epoch[011] Batch [3249]/[3759]	Speed: 70.780476 samples/sec	 accuracy=47.327885	 loss=2.256663	 lr=0.010000
Epoch[011] Batch [3299]/[3759]	Speed: 68.767016 samples/sec	 accuracy=47.345644	 loss=2.256714	 lr=0.010000
Epoch[011] Batch [3349]/[3759]	Speed: 67.565006 samples/sec	 accuracy=47.330224	 loss=2.257312	 lr=0.010000
Epoch[011] Batch [3399]/[3759]	Speed: 71.566754 samples/sec	 accuracy=47.332261	 loss=2.257381	 lr=0.010000
Epoch[011] Batch [3449]/[3759]	Speed: 69.433369 samples/sec	 accuracy=47.299366	 loss=2.258903	 lr=0.010000
Epoch[011] Batch [3499]/[3759]	Speed: 68.101660 samples/sec	 accuracy=47.275893	 loss=2.260154	 lr=0.010000
Epoch[011] Batch [3549]/[3759]	Speed: 69.973454 samples/sec	 accuracy=47.274648	 loss=2.260293	 lr=0.010000
Epoch[011] Batch [3599]/[3759]	Speed: 70.408120 samples/sec	 accuracy=47.281684	 loss=2.259980	 lr=0.010000
Epoch[011] Batch [3649]/[3759]	Speed: 69.207257 samples/sec	 accuracy=47.278682	 loss=2.260972	 lr=0.010000
Epoch[011] Batch [3699]/[3759]	Speed: 69.288441 samples/sec	 accuracy=47.282095	 loss=2.261000	 lr=0.010000
Epoch[011] Batch [3749]/[3759]	Speed: 77.323156 samples/sec	 accuracy=47.270833	 loss=2.261086	 lr=0.010000
Batch [0049]/[0303]: acc-top1=48.937500 acc-top5=75.031250
Batch [0099]/[0303]: acc-top1=48.578125 acc-top5=74.187500
Batch [0149]/[0303]: acc-top1=49.187500 acc-top5=74.906250
Batch [0199]/[0303]: acc-top1=49.484375 acc-top5=75.218750
Batch [0249]/[0303]: acc-top1=49.512500 acc-top5=75.243750
Batch [0299]/[0303]: acc-top1=49.442708 acc-top5=74.932292
[Epoch 011] training: accuracy=47.269886	 loss=2.261113
[Epoch 011] speed: 69 samples/sec	time cost: 3791.915097
[Epoch 011] validation: acc-top1=49.437913 acc-top5=74.876238 loss=2.303145
Epoch[012] Batch [0049]/[3760]	Speed: 48.290884 samples/sec	 accuracy=47.625000	 loss=2.243705	 lr=0.010000
Epoch[012] Batch [0099]/[3760]	Speed: 69.520273 samples/sec	 accuracy=47.843750	 loss=2.214010	 lr=0.010000
Epoch[012] Batch [0149]/[3760]	Speed: 67.506284 samples/sec	 accuracy=48.635417	 loss=2.187747	 lr=0.010000
Epoch[012] Batch [0199]/[3760]	Speed: 69.935439 samples/sec	 accuracy=48.234375	 loss=2.213434	 lr=0.010000
Epoch[012] Batch [0249]/[3760]	Speed: 67.936783 samples/sec	 accuracy=48.281250	 loss=2.223307	 lr=0.010000
Epoch[012] Batch [0299]/[3760]	Speed: 70.229956 samples/sec	 accuracy=48.192708	 loss=2.227942	 lr=0.010000
Epoch[012] Batch [0349]/[3760]	Speed: 68.640315 samples/sec	 accuracy=48.419643	 loss=2.220042	 lr=0.010000
Epoch[012] Batch [0399]/[3760]	Speed: 71.288365 samples/sec	 accuracy=48.472656	 loss=2.219431	 lr=0.010000
Epoch[012] Batch [0449]/[3760]	Speed: 70.785073 samples/sec	 accuracy=48.322917	 loss=2.225755	 lr=0.010000
Epoch[012] Batch [0499]/[3760]	Speed: 68.311379 samples/sec	 accuracy=48.293750	 loss=2.228625	 lr=0.010000
Epoch[012] Batch [0549]/[3760]	Speed: 70.113236 samples/sec	 accuracy=48.400568	 loss=2.223330	 lr=0.010000
Epoch[012] Batch [0599]/[3760]	Speed: 69.848112 samples/sec	 accuracy=48.500000	 loss=2.219778	 lr=0.010000
Epoch[012] Batch [0649]/[3760]	Speed: 68.036705 samples/sec	 accuracy=48.403846	 loss=2.221667	 lr=0.010000
Epoch[012] Batch [0699]/[3760]	Speed: 70.207464 samples/sec	 accuracy=48.287946	 loss=2.224929	 lr=0.010000
Epoch[012] Batch [0749]/[3760]	Speed: 65.940050 samples/sec	 accuracy=48.114583	 loss=2.231663	 lr=0.010000
Epoch[012] Batch [0799]/[3760]	Speed: 72.961104 samples/sec	 accuracy=48.126953	 loss=2.231099	 lr=0.010000
Epoch[012] Batch [0849]/[3760]	Speed: 70.469575 samples/sec	 accuracy=48.141544	 loss=2.228572	 lr=0.010000
Epoch[012] Batch [0899]/[3760]	Speed: 69.701839 samples/sec	 accuracy=48.166667	 loss=2.226528	 lr=0.010000
Epoch[012] Batch [0949]/[3760]	Speed: 67.620133 samples/sec	 accuracy=48.108553	 loss=2.227925	 lr=0.010000
Epoch[012] Batch [0999]/[3760]	Speed: 68.965526 samples/sec	 accuracy=48.148438	 loss=2.228507	 lr=0.010000
Epoch[012] Batch [1049]/[3760]	Speed: 70.887049 samples/sec	 accuracy=48.116071	 loss=2.232716	 lr=0.010000
Epoch[012] Batch [1099]/[3760]	Speed: 68.133636 samples/sec	 accuracy=48.146307	 loss=2.230703	 lr=0.010000
Epoch[012] Batch [1149]/[3760]	Speed: 68.965192 samples/sec	 accuracy=48.150815	 loss=2.229062	 lr=0.010000
Epoch[012] Batch [1199]/[3760]	Speed: 69.860241 samples/sec	 accuracy=48.119792	 loss=2.229201	 lr=0.010000
Epoch[012] Batch [1249]/[3760]	Speed: 69.056550 samples/sec	 accuracy=48.063750	 loss=2.230979	 lr=0.010000
Epoch[012] Batch [1299]/[3760]	Speed: 71.032841 samples/sec	 accuracy=48.026442	 loss=2.231476	 lr=0.010000
Epoch[012] Batch [1349]/[3760]	Speed: 68.578777 samples/sec	 accuracy=48.049769	 loss=2.231432	 lr=0.010000
Epoch[012] Batch [1399]/[3760]	Speed: 70.470622 samples/sec	 accuracy=48.032366	 loss=2.231963	 lr=0.010000
Epoch[012] Batch [1449]/[3760]	Speed: 70.132124 samples/sec	 accuracy=48.059267	 loss=2.230668	 lr=0.010000
Epoch[012] Batch [1499]/[3760]	Speed: 68.316084 samples/sec	 accuracy=48.083333	 loss=2.228079	 lr=0.010000
Epoch[012] Batch [1549]/[3760]	Speed: 70.516011 samples/sec	 accuracy=48.095766	 loss=2.227542	 lr=0.010000
Epoch[012] Batch [1599]/[3760]	Speed: 67.446323 samples/sec	 accuracy=48.064453	 loss=2.228966	 lr=0.010000
Epoch[012] Batch [1649]/[3760]	Speed: 70.930616 samples/sec	 accuracy=48.079545	 loss=2.227330	 lr=0.010000
Epoch[012] Batch [1699]/[3760]	Speed: 69.424994 samples/sec	 accuracy=48.082721	 loss=2.228327	 lr=0.010000
Epoch[012] Batch [1749]/[3760]	Speed: 68.989167 samples/sec	 accuracy=48.065179	 loss=2.229227	 lr=0.010000
Epoch[012] Batch [1799]/[3760]	Speed: 70.303325 samples/sec	 accuracy=48.059896	 loss=2.229132	 lr=0.010000
Epoch[012] Batch [1849]/[3760]	Speed: 66.712844 samples/sec	 accuracy=48.081081	 loss=2.228841	 lr=0.010000
Epoch[012] Batch [1899]/[3760]	Speed: 70.861704 samples/sec	 accuracy=48.077303	 loss=2.229140	 lr=0.010000
Epoch[012] Batch [1949]/[3760]	Speed: 68.492163 samples/sec	 accuracy=48.068910	 loss=2.227827	 lr=0.010000
Epoch[012] Batch [1999]/[3760]	Speed: 69.679397 samples/sec	 accuracy=48.104687	 loss=2.226321	 lr=0.010000
Epoch[012] Batch [2049]/[3760]	Speed: 71.392641 samples/sec	 accuracy=48.111280	 loss=2.226737	 lr=0.010000
Epoch[012] Batch [2099]/[3760]	Speed: 68.462061 samples/sec	 accuracy=48.180804	 loss=2.224871	 lr=0.010000
Epoch[012] Batch [2149]/[3760]	Speed: 68.027211 samples/sec	 accuracy=48.220203	 loss=2.223221	 lr=0.010000
Epoch[012] Batch [2199]/[3760]	Speed: 68.621242 samples/sec	 accuracy=48.229403	 loss=2.223622	 lr=0.010000
Epoch[012] Batch [2249]/[3760]	Speed: 71.126038 samples/sec	 accuracy=48.235417	 loss=2.223784	 lr=0.010000
Epoch[012] Batch [2299]/[3760]	Speed: 70.996913 samples/sec	 accuracy=48.197690	 loss=2.225318	 lr=0.010000
Epoch[012] Batch [2349]/[3760]	Speed: 67.425384 samples/sec	 accuracy=48.208112	 loss=2.224719	 lr=0.010000
Epoch[012] Batch [2399]/[3760]	Speed: 70.808688 samples/sec	 accuracy=48.172526	 loss=2.225202	 lr=0.010000
Epoch[012] Batch [2449]/[3760]	Speed: 67.100062 samples/sec	 accuracy=48.160077	 loss=2.225886	 lr=0.010000
Epoch[012] Batch [2499]/[3760]	Speed: 70.979699 samples/sec	 accuracy=48.150000	 loss=2.227141	 lr=0.010000
Epoch[012] Batch [2549]/[3760]	Speed: 63.613266 samples/sec	 accuracy=48.159926	 loss=2.225654	 lr=0.010000
Epoch[012] Batch [2599]/[3760]	Speed: 72.565322 samples/sec	 accuracy=48.148438	 loss=2.225716	 lr=0.010000
Epoch[012] Batch [2649]/[3760]	Speed: 71.181243 samples/sec	 accuracy=48.130896	 loss=2.224997	 lr=0.010000
Epoch[012] Batch [2699]/[3760]	Speed: 64.799771 samples/sec	 accuracy=48.161458	 loss=2.224406	 lr=0.010000
Epoch[012] Batch [2749]/[3760]	Speed: 72.787624 samples/sec	 accuracy=48.151705	 loss=2.224857	 lr=0.010000
Epoch[012] Batch [2799]/[3760]	Speed: 69.219522 samples/sec	 accuracy=48.141741	 loss=2.225110	 lr=0.010000
Epoch[012] Batch [2849]/[3760]	Speed: 70.969156 samples/sec	 accuracy=48.163377	 loss=2.225634	 lr=0.010000
Epoch[012] Batch [2899]/[3760]	Speed: 66.770564 samples/sec	 accuracy=48.149784	 loss=2.225610	 lr=0.010000
Epoch[012] Batch [2949]/[3760]	Speed: 69.643439 samples/sec	 accuracy=48.147775	 loss=2.225506	 lr=0.010000
Epoch[012] Batch [2999]/[3760]	Speed: 71.623254 samples/sec	 accuracy=48.153125	 loss=2.226151	 lr=0.010000
Epoch[012] Batch [3049]/[3760]	Speed: 68.674709 samples/sec	 accuracy=48.132172	 loss=2.227673	 lr=0.010000
Epoch[012] Batch [3099]/[3760]	Speed: 69.243857 samples/sec	 accuracy=48.129032	 loss=2.227433	 lr=0.010000
Epoch[012] Batch [3149]/[3760]	Speed: 69.371244 samples/sec	 accuracy=48.122024	 loss=2.227516	 lr=0.010000
Epoch[012] Batch [3199]/[3760]	Speed: 67.545655 samples/sec	 accuracy=48.113770	 loss=2.227096	 lr=0.010000
Epoch[012] Batch [3249]/[3760]	Speed: 70.928273 samples/sec	 accuracy=48.103365	 loss=2.227350	 lr=0.010000
Epoch[012] Batch [3299]/[3760]	Speed: 69.945853 samples/sec	 accuracy=48.102273	 loss=2.227859	 lr=0.010000
Epoch[012] Batch [3349]/[3760]	Speed: 69.591975 samples/sec	 accuracy=48.098881	 loss=2.227428	 lr=0.010000
Epoch[012] Batch [3399]/[3760]	Speed: 68.855402 samples/sec	 accuracy=48.086857	 loss=2.227675	 lr=0.010000
Epoch[012] Batch [3449]/[3760]	Speed: 68.492181 samples/sec	 accuracy=48.068841	 loss=2.228527	 lr=0.010000
Epoch[012] Batch [3499]/[3760]	Speed: 69.964641 samples/sec	 accuracy=48.077232	 loss=2.228390	 lr=0.010000
Epoch[012] Batch [3549]/[3760]	Speed: 68.645469 samples/sec	 accuracy=48.072623	 loss=2.228695	 lr=0.010000
Epoch[012] Batch [3599]/[3760]	Speed: 70.126952 samples/sec	 accuracy=48.075955	 loss=2.228412	 lr=0.010000
Epoch[012] Batch [3649]/[3760]	Speed: 69.299338 samples/sec	 accuracy=48.081764	 loss=2.227921	 lr=0.010000
Epoch[012] Batch [3699]/[3760]	Speed: 69.815536 samples/sec	 accuracy=48.079392	 loss=2.227311	 lr=0.010000
Epoch[012] Batch [3749]/[3760]	Speed: 77.143746 samples/sec	 accuracy=48.083333	 loss=2.226897	 lr=0.010000
Batch [0049]/[0303]: acc-top1=49.031250 acc-top5=75.468750
Batch [0099]/[0303]: acc-top1=49.437500 acc-top5=75.375000
Batch [0149]/[0303]: acc-top1=49.906250 acc-top5=75.656250
Batch [0199]/[0303]: acc-top1=49.953125 acc-top5=75.914062
Batch [0249]/[0303]: acc-top1=50.031250 acc-top5=76.118750
Batch [0299]/[0303]: acc-top1=49.786458 acc-top5=75.864583
[Epoch 012] training: accuracy=48.082197	 loss=2.227053
[Epoch 012] speed: 69 samples/sec	time cost: 3786.765672
[Epoch 012] validation: acc-top1=49.762789 acc-top5=75.835396 loss=2.248447
Epoch[013] Batch [0049]/[3760]	Speed: 47.884078 samples/sec	 accuracy=49.000000	 loss=2.194573	 lr=0.010000
Epoch[013] Batch [0099]/[3760]	Speed: 69.108358 samples/sec	 accuracy=48.968750	 loss=2.175473	 lr=0.010000
Epoch[013] Batch [0149]/[3760]	Speed: 68.694200 samples/sec	 accuracy=48.739583	 loss=2.189534	 lr=0.010000
Epoch[013] Batch [0199]/[3760]	Speed: 70.806525 samples/sec	 accuracy=48.632812	 loss=2.188809	 lr=0.010000
Epoch[013] Batch [0249]/[3760]	Speed: 65.915975 samples/sec	 accuracy=48.875000	 loss=2.182144	 lr=0.010000
Epoch[013] Batch [0299]/[3760]	Speed: 71.761859 samples/sec	 accuracy=48.713542	 loss=2.188029	 lr=0.010000
Epoch[013] Batch [0349]/[3760]	Speed: 70.144278 samples/sec	 accuracy=48.830357	 loss=2.183624	 lr=0.010000
Epoch[013] Batch [0399]/[3760]	Speed: 71.143210 samples/sec	 accuracy=48.933594	 loss=2.182464	 lr=0.010000
Epoch[013] Batch [0449]/[3760]	Speed: 69.305160 samples/sec	 accuracy=48.888889	 loss=2.182411	 lr=0.010000
Epoch[013] Batch [0499]/[3760]	Speed: 70.054889 samples/sec	 accuracy=48.762500	 loss=2.189206	 lr=0.010000
Epoch[013] Batch [0549]/[3760]	Speed: 68.446645 samples/sec	 accuracy=48.755682	 loss=2.188959	 lr=0.010000
Epoch[013] Batch [0599]/[3760]	Speed: 70.777970 samples/sec	 accuracy=48.682292	 loss=2.191284	 lr=0.010000
Epoch[013] Batch [0649]/[3760]	Speed: 69.877807 samples/sec	 accuracy=48.637019	 loss=2.191490	 lr=0.010000
Epoch[013] Batch [0699]/[3760]	Speed: 69.357105 samples/sec	 accuracy=48.598214	 loss=2.193186	 lr=0.010000
Epoch[013] Batch [0749]/[3760]	Speed: 70.439464 samples/sec	 accuracy=48.625000	 loss=2.194010	 lr=0.010000
Epoch[013] Batch [0799]/[3760]	Speed: 69.158083 samples/sec	 accuracy=48.753906	 loss=2.193528	 lr=0.010000
Epoch[013] Batch [0849]/[3760]	Speed: 69.845830 samples/sec	 accuracy=48.781250	 loss=2.192378	 lr=0.010000
Epoch[013] Batch [0899]/[3760]	Speed: 67.808549 samples/sec	 accuracy=48.798611	 loss=2.194511	 lr=0.010000
Epoch[013] Batch [0949]/[3760]	Speed: 71.088035 samples/sec	 accuracy=48.720395	 loss=2.198797	 lr=0.010000
Epoch[013] Batch [0999]/[3760]	Speed: 70.252945 samples/sec	 accuracy=48.715625	 loss=2.198575	 lr=0.010000
Epoch[013] Batch [1049]/[3760]	Speed: 70.046379 samples/sec	 accuracy=48.678571	 loss=2.197712	 lr=0.010000
Epoch[013] Batch [1099]/[3760]	Speed: 70.227768 samples/sec	 accuracy=48.690341	 loss=2.196780	 lr=0.010000
Epoch[013] Batch [1149]/[3760]	Speed: 68.690960 samples/sec	 accuracy=48.775815	 loss=2.192956	 lr=0.010000
Epoch[013] Batch [1199]/[3760]	Speed: 69.578932 samples/sec	 accuracy=48.803385	 loss=2.194055	 lr=0.010000
Epoch[013] Batch [1249]/[3760]	Speed: 69.528549 samples/sec	 accuracy=48.733750	 loss=2.195564	 lr=0.010000
Epoch[013] Batch [1299]/[3760]	Speed: 69.024005 samples/sec	 accuracy=48.685096	 loss=2.198503	 lr=0.010000
Epoch[013] Batch [1349]/[3760]	Speed: 70.243095 samples/sec	 accuracy=48.776620	 loss=2.194880	 lr=0.010000
Epoch[013] Batch [1399]/[3760]	Speed: 69.030820 samples/sec	 accuracy=48.821429	 loss=2.194079	 lr=0.010000
Epoch[013] Batch [1449]/[3760]	Speed: 70.495543 samples/sec	 accuracy=48.823276	 loss=2.193071	 lr=0.010000
Epoch[013] Batch [1499]/[3760]	Speed: 69.297202 samples/sec	 accuracy=48.812500	 loss=2.192901	 lr=0.010000
Epoch[013] Batch [1549]/[3760]	Speed: 69.594219 samples/sec	 accuracy=48.812500	 loss=2.191425	 lr=0.010000
Epoch[013] Batch [1599]/[3760]	Speed: 68.243763 samples/sec	 accuracy=48.794922	 loss=2.192937	 lr=0.010000
Epoch[013] Batch [1649]/[3760]	Speed: 71.667273 samples/sec	 accuracy=48.822917	 loss=2.193163	 lr=0.010000
Epoch[013] Batch [1699]/[3760]	Speed: 68.348944 samples/sec	 accuracy=48.781250	 loss=2.193840	 lr=0.010000
Epoch[013] Batch [1749]/[3760]	Speed: 70.242291 samples/sec	 accuracy=48.784821	 loss=2.194187	 lr=0.010000
Epoch[013] Batch [1799]/[3760]	Speed: 69.815076 samples/sec	 accuracy=48.733507	 loss=2.195075	 lr=0.010000
Epoch[013] Batch [1849]/[3760]	Speed: 70.838701 samples/sec	 accuracy=48.728885	 loss=2.195445	 lr=0.010000
Epoch[013] Batch [1899]/[3760]	Speed: 69.075115 samples/sec	 accuracy=48.710526	 loss=2.196578	 lr=0.010000
Epoch[013] Batch [1949]/[3760]	Speed: 68.451967 samples/sec	 accuracy=48.701122	 loss=2.196090	 lr=0.010000
Epoch[013] Batch [1999]/[3760]	Speed: 70.729037 samples/sec	 accuracy=48.683594	 loss=2.196244	 lr=0.010000
Epoch[013] Batch [2049]/[3760]	Speed: 69.455651 samples/sec	 accuracy=48.682927	 loss=2.197057	 lr=0.010000
Epoch[013] Batch [2099]/[3760]	Speed: 69.655058 samples/sec	 accuracy=48.646577	 loss=2.197889	 lr=0.010000
Epoch[013] Batch [2149]/[3760]	Speed: 68.384567 samples/sec	 accuracy=48.642442	 loss=2.197551	 lr=0.010000
Epoch[013] Batch [2199]/[3760]	Speed: 71.178179 samples/sec	 accuracy=48.646307	 loss=2.197511	 lr=0.010000
Epoch[013] Batch [2249]/[3760]	Speed: 68.821732 samples/sec	 accuracy=48.661111	 loss=2.196712	 lr=0.010000
Epoch[013] Batch [2299]/[3760]	Speed: 69.365160 samples/sec	 accuracy=48.653533	 loss=2.196723	 lr=0.010000
Epoch[013] Batch [2349]/[3760]	Speed: 68.514931 samples/sec	 accuracy=48.596410	 loss=2.198457	 lr=0.010000
Epoch[013] Batch [2399]/[3760]	Speed: 71.062364 samples/sec	 accuracy=48.604167	 loss=2.198129	 lr=0.010000
Epoch[013] Batch [2449]/[3760]	Speed: 70.036279 samples/sec	 accuracy=48.593750	 loss=2.198447	 lr=0.010000
Epoch[013] Batch [2499]/[3760]	Speed: 69.725646 samples/sec	 accuracy=48.612500	 loss=2.197441	 lr=0.010000
Epoch[013] Batch [2549]/[3760]	Speed: 68.868421 samples/sec	 accuracy=48.602941	 loss=2.198728	 lr=0.010000
Epoch[013] Batch [2599]/[3760]	Speed: 69.301101 samples/sec	 accuracy=48.615986	 loss=2.198148	 lr=0.010000
Epoch[013] Batch [2649]/[3760]	Speed: 70.023363 samples/sec	 accuracy=48.606132	 loss=2.198006	 lr=0.010000
Epoch[013] Batch [2699]/[3760]	Speed: 69.434924 samples/sec	 accuracy=48.637153	 loss=2.196707	 lr=0.010000
Epoch[013] Batch [2749]/[3760]	Speed: 69.286456 samples/sec	 accuracy=48.647727	 loss=2.197476	 lr=0.010000
Epoch[013] Batch [2799]/[3760]	Speed: 69.385561 samples/sec	 accuracy=48.612723	 loss=2.198516	 lr=0.010000
Epoch[013] Batch [2849]/[3760]	Speed: 71.493016 samples/sec	 accuracy=48.623904	 loss=2.198296	 lr=0.010000
Epoch[013] Batch [2899]/[3760]	Speed: 68.049048 samples/sec	 accuracy=48.593750	 loss=2.198967	 lr=0.010000
Epoch[013] Batch [2949]/[3760]	Speed: 69.228385 samples/sec	 accuracy=48.572564	 loss=2.199223	 lr=0.010000
Epoch[013] Batch [2999]/[3760]	Speed: 71.534081 samples/sec	 accuracy=48.583333	 loss=2.198810	 lr=0.010000
Epoch[013] Batch [3049]/[3760]	Speed: 67.984987 samples/sec	 accuracy=48.605533	 loss=2.198433	 lr=0.010000
Epoch[013] Batch [3099]/[3760]	Speed: 70.296749 samples/sec	 accuracy=48.609879	 loss=2.197821	 lr=0.010000
Epoch[013] Batch [3149]/[3760]	Speed: 68.302733 samples/sec	 accuracy=48.604167	 loss=2.197634	 lr=0.010000
Epoch[013] Batch [3199]/[3760]	Speed: 70.917795 samples/sec	 accuracy=48.598145	 loss=2.197394	 lr=0.010000
Epoch[013] Batch [3249]/[3760]	Speed: 67.802729 samples/sec	 accuracy=48.611058	 loss=2.196472	 lr=0.010000
Epoch[013] Batch [3299]/[3760]	Speed: 71.588293 samples/sec	 accuracy=48.575284	 loss=2.198068	 lr=0.010000
Epoch[013] Batch [3349]/[3760]	Speed: 68.523584 samples/sec	 accuracy=48.550840	 loss=2.199627	 lr=0.010000
Epoch[013] Batch [3399]/[3760]	Speed: 69.816034 samples/sec	 accuracy=48.559743	 loss=2.199260	 lr=0.010000
Epoch[013] Batch [3449]/[3760]	Speed: 70.405650 samples/sec	 accuracy=48.552083	 loss=2.199822	 lr=0.010000
Epoch[013] Batch [3499]/[3760]	Speed: 69.978753 samples/sec	 accuracy=48.514732	 loss=2.200121	 lr=0.010000
Epoch[013] Batch [3549]/[3760]	Speed: 70.836993 samples/sec	 accuracy=48.500000	 loss=2.200999	 lr=0.010000
Epoch[013] Batch [3599]/[3760]	Speed: 68.023888 samples/sec	 accuracy=48.519965	 loss=2.200300	 lr=0.010000
Epoch[013] Batch [3649]/[3760]	Speed: 70.872749 samples/sec	 accuracy=48.518836	 loss=2.200384	 lr=0.010000
Epoch[013] Batch [3699]/[3760]	Speed: 68.832698 samples/sec	 accuracy=48.543074	 loss=2.199044	 lr=0.010000
Epoch[013] Batch [3749]/[3760]	Speed: 78.641376 samples/sec	 accuracy=48.535417	 loss=2.199589	 lr=0.010000
Batch [0049]/[0303]: acc-top1=50.718750 acc-top5=75.562500
Batch [0099]/[0303]: acc-top1=50.250000 acc-top5=75.281250
Batch [0149]/[0303]: acc-top1=50.395833 acc-top5=75.958333
Batch [0199]/[0303]: acc-top1=50.500000 acc-top5=76.117188
Batch [0249]/[0303]: acc-top1=50.693750 acc-top5=76.375000
Batch [0299]/[0303]: acc-top1=50.557292 acc-top5=76.223958
[Epoch 013] training: accuracy=48.533910	 loss=2.199480
[Epoch 013] speed: 69 samples/sec	time cost: 3769.272954
[Epoch 013] validation: acc-top1=50.562087 acc-top5=76.191213 loss=2.191052
Epoch[014] Batch [0049]/[3759]	Speed: 48.358251 samples/sec	 accuracy=50.218750	 loss=2.093772	 lr=0.010000
Epoch[014] Batch [0099]/[3759]	Speed: 67.441073 samples/sec	 accuracy=49.937500	 loss=2.107116	 lr=0.010000
Epoch[014] Batch [0149]/[3759]	Speed: 69.318509 samples/sec	 accuracy=49.645833	 loss=2.131253	 lr=0.010000
Epoch[014] Batch [0199]/[3759]	Speed: 69.020279 samples/sec	 accuracy=49.203125	 loss=2.144759	 lr=0.010000
Epoch[014] Batch [0249]/[3759]	Speed: 69.724353 samples/sec	 accuracy=49.168750	 loss=2.148720	 lr=0.010000
Epoch[014] Batch [0299]/[3759]	Speed: 70.431035 samples/sec	 accuracy=49.145833	 loss=2.148132	 lr=0.010000
Epoch[014] Batch [0349]/[3759]	Speed: 68.337270 samples/sec	 accuracy=49.125000	 loss=2.151524	 lr=0.010000
Epoch[014] Batch [0399]/[3759]	Speed: 71.143718 samples/sec	 accuracy=49.007812	 loss=2.153796	 lr=0.010000
Epoch[014] Batch [0449]/[3759]	Speed: 69.929151 samples/sec	 accuracy=49.024306	 loss=2.149097	 lr=0.010000
Epoch[014] Batch [0499]/[3759]	Speed: 68.318448 samples/sec	 accuracy=48.971875	 loss=2.150716	 lr=0.010000
Epoch[014] Batch [0549]/[3759]	Speed: 69.468491 samples/sec	 accuracy=49.014205	 loss=2.151246	 lr=0.010000
Epoch[014] Batch [0599]/[3759]	Speed: 68.656665 samples/sec	 accuracy=49.111979	 loss=2.152418	 lr=0.010000
Epoch[014] Batch [0649]/[3759]	Speed: 71.212247 samples/sec	 accuracy=49.103365	 loss=2.150480	 lr=0.010000
Epoch[014] Batch [0699]/[3759]	Speed: 68.694194 samples/sec	 accuracy=49.120536	 loss=2.150270	 lr=0.010000
Epoch[014] Batch [0749]/[3759]	Speed: 68.768320 samples/sec	 accuracy=49.258333	 loss=2.149773	 lr=0.010000
Epoch[014] Batch [0799]/[3759]	Speed: 70.276822 samples/sec	 accuracy=49.179688	 loss=2.151795	 lr=0.010000
Epoch[014] Batch [0849]/[3759]	Speed: 69.421415 samples/sec	 accuracy=49.207721	 loss=2.152566	 lr=0.010000
Epoch[014] Batch [0899]/[3759]	Speed: 67.192195 samples/sec	 accuracy=49.229167	 loss=2.152869	 lr=0.010000
Epoch[014] Batch [0949]/[3759]	Speed: 71.666128 samples/sec	 accuracy=49.223684	 loss=2.155848	 lr=0.010000
Epoch[014] Batch [0999]/[3759]	Speed: 70.180131 samples/sec	 accuracy=49.223438	 loss=2.156702	 lr=0.010000
Epoch[014] Batch [1049]/[3759]	Speed: 70.101551 samples/sec	 accuracy=49.270833	 loss=2.158168	 lr=0.010000
Epoch[014] Batch [1099]/[3759]	Speed: 69.106379 samples/sec	 accuracy=49.262784	 loss=2.159701	 lr=0.010000
Epoch[014] Batch [1149]/[3759]	Speed: 69.872773 samples/sec	 accuracy=49.319293	 loss=2.158636	 lr=0.010000
Epoch[014] Batch [1199]/[3759]	Speed: 68.924532 samples/sec	 accuracy=49.320312	 loss=2.158457	 lr=0.010000
Epoch[014] Batch [1249]/[3759]	Speed: 68.924448 samples/sec	 accuracy=49.342500	 loss=2.158023	 lr=0.010000
Epoch[014] Batch [1299]/[3759]	Speed: 69.477191 samples/sec	 accuracy=49.302885	 loss=2.158994	 lr=0.010000
Epoch[014] Batch [1349]/[3759]	Speed: 70.303635 samples/sec	 accuracy=49.306713	 loss=2.159179	 lr=0.010000
Epoch[014] Batch [1399]/[3759]	Speed: 67.692565 samples/sec	 accuracy=49.367188	 loss=2.158763	 lr=0.010000
Epoch[014] Batch [1449]/[3759]	Speed: 70.690167 samples/sec	 accuracy=49.337284	 loss=2.158460	 lr=0.010000
Epoch[014] Batch [1499]/[3759]	Speed: 70.917149 samples/sec	 accuracy=49.335417	 loss=2.158909	 lr=0.010000
Epoch[014] Batch [1549]/[3759]	Speed: 69.664355 samples/sec	 accuracy=49.360887	 loss=2.158130	 lr=0.010000
Epoch[014] Batch [1599]/[3759]	Speed: 68.784297 samples/sec	 accuracy=49.326172	 loss=2.158804	 lr=0.010000
Epoch[014] Batch [1649]/[3759]	Speed: 69.289558 samples/sec	 accuracy=49.310606	 loss=2.160411	 lr=0.010000
Epoch[014] Batch [1699]/[3759]	Speed: 71.022817 samples/sec	 accuracy=49.302390	 loss=2.161285	 lr=0.010000
Epoch[014] Batch [1749]/[3759]	Speed: 67.407444 samples/sec	 accuracy=49.327679	 loss=2.161421	 lr=0.010000
Epoch[014] Batch [1799]/[3759]	Speed: 71.582473 samples/sec	 accuracy=49.313368	 loss=2.162833	 lr=0.010000
Epoch[014] Batch [1849]/[3759]	Speed: 70.249577 samples/sec	 accuracy=49.334459	 loss=2.161988	 lr=0.010000
Epoch[014] Batch [1899]/[3759]	Speed: 69.437867 samples/sec	 accuracy=49.301809	 loss=2.162056	 lr=0.010000
Epoch[014] Batch [1949]/[3759]	Speed: 67.986253 samples/sec	 accuracy=49.294872	 loss=2.162692	 lr=0.010000
Epoch[014] Batch [1999]/[3759]	Speed: 70.405830 samples/sec	 accuracy=49.295313	 loss=2.162933	 lr=0.010000
Epoch[014] Batch [2049]/[3759]	Speed: 69.764985 samples/sec	 accuracy=49.307165	 loss=2.163409	 lr=0.010000
Epoch[014] Batch [2099]/[3759]	Speed: 70.247352 samples/sec	 accuracy=49.316964	 loss=2.163490	 lr=0.010000
Epoch[014] Batch [2149]/[3759]	Speed: 69.643161 samples/sec	 accuracy=49.311773	 loss=2.163871	 lr=0.010000
Epoch[014] Batch [2199]/[3759]	Speed: 69.828872 samples/sec	 accuracy=49.343750	 loss=2.163713	 lr=0.010000
Epoch[014] Batch [2249]/[3759]	Speed: 70.561683 samples/sec	 accuracy=49.313889	 loss=2.165225	 lr=0.010000
Epoch[014] Batch [2299]/[3759]	Speed: 68.642922 samples/sec	 accuracy=49.306386	 loss=2.166272	 lr=0.010000
Epoch[014] Batch [2349]/[3759]	Speed: 67.821469 samples/sec	 accuracy=49.255319	 loss=2.167920	 lr=0.010000
Epoch[014] Batch [2399]/[3759]	Speed: 71.561983 samples/sec	 accuracy=49.228516	 loss=2.169106	 lr=0.010000
Epoch[014] Batch [2449]/[3759]	Speed: 70.497591 samples/sec	 accuracy=49.260204	 loss=2.168622	 lr=0.010000
Epoch[014] Batch [2499]/[3759]	Speed: 69.219961 samples/sec	 accuracy=49.272500	 loss=2.168334	 lr=0.010000
Epoch[014] Batch [2549]/[3759]	Speed: 70.142685 samples/sec	 accuracy=49.293505	 loss=2.167803	 lr=0.010000
Epoch[014] Batch [2599]/[3759]	Speed: 68.378457 samples/sec	 accuracy=49.290264	 loss=2.167983	 lr=0.010000
Epoch[014] Batch [2649]/[3759]	Speed: 69.418817 samples/sec	 accuracy=49.271226	 loss=2.168839	 lr=0.010000
Epoch[014] Batch [2699]/[3759]	Speed: 70.827635 samples/sec	 accuracy=49.289931	 loss=2.169309	 lr=0.010000
Epoch[014] Batch [2749]/[3759]	Speed: 67.910967 samples/sec	 accuracy=49.258523	 loss=2.170691	 lr=0.010000
Epoch[014] Batch [2799]/[3759]	Speed: 70.908143 samples/sec	 accuracy=49.232701	 loss=2.170571	 lr=0.010000
Epoch[014] Batch [2849]/[3759]	Speed: 70.412793 samples/sec	 accuracy=49.250000	 loss=2.169703	 lr=0.010000
Epoch[014] Batch [2899]/[3759]	Speed: 68.367331 samples/sec	 accuracy=49.252694	 loss=2.170069	 lr=0.010000
Epoch[014] Batch [2949]/[3759]	Speed: 70.287614 samples/sec	 accuracy=49.224576	 loss=2.170819	 lr=0.010000
Epoch[014] Batch [2999]/[3759]	Speed: 70.910589 samples/sec	 accuracy=49.216146	 loss=2.171306	 lr=0.010000
Epoch[014] Batch [3049]/[3759]	Speed: 70.003109 samples/sec	 accuracy=49.188012	 loss=2.171888	 lr=0.010000
Epoch[014] Batch [3099]/[3759]	Speed: 68.938074 samples/sec	 accuracy=49.171371	 loss=2.172426	 lr=0.010000
Epoch[014] Batch [3149]/[3759]	Speed: 69.494536 samples/sec	 accuracy=49.172123	 loss=2.172863	 lr=0.010000
Epoch[014] Batch [3199]/[3759]	Speed: 70.226343 samples/sec	 accuracy=49.175293	 loss=2.172386	 lr=0.010000
Epoch[014] Batch [3249]/[3759]	Speed: 70.212261 samples/sec	 accuracy=49.158173	 loss=2.172845	 lr=0.010000
Epoch[014] Batch [3299]/[3759]	Speed: 67.623107 samples/sec	 accuracy=49.151042	 loss=2.173591	 lr=0.010000
Epoch[014] Batch [3349]/[3759]	Speed: 71.013031 samples/sec	 accuracy=49.131530	 loss=2.173966	 lr=0.010000
Epoch[014] Batch [3399]/[3759]	Speed: 69.780192 samples/sec	 accuracy=49.100184	 loss=2.174611	 lr=0.010000
Epoch[014] Batch [3449]/[3759]	Speed: 68.298811 samples/sec	 accuracy=49.093750	 loss=2.174543	 lr=0.010000
Epoch[014] Batch [3499]/[3759]	Speed: 70.280722 samples/sec	 accuracy=49.080804	 loss=2.174895	 lr=0.010000
Epoch[014] Batch [3549]/[3759]	Speed: 70.478129 samples/sec	 accuracy=49.081866	 loss=2.175369	 lr=0.010000
Epoch[014] Batch [3599]/[3759]	Speed: 69.829259 samples/sec	 accuracy=49.081597	 loss=2.175599	 lr=0.010000
Epoch[014] Batch [3649]/[3759]	Speed: 68.389224 samples/sec	 accuracy=49.074914	 loss=2.175347	 lr=0.010000
Epoch[014] Batch [3699]/[3759]	Speed: 70.530145 samples/sec	 accuracy=49.073902	 loss=2.175176	 lr=0.010000
Epoch[014] Batch [3749]/[3759]	Speed: 77.613572 samples/sec	 accuracy=49.057500	 loss=2.175501	 lr=0.010000
Batch [0049]/[0303]: acc-top1=50.281250 acc-top5=76.250000
Batch [0099]/[0303]: acc-top1=50.656250 acc-top5=76.140625
Batch [0149]/[0303]: acc-top1=50.854167 acc-top5=76.395833
Batch [0199]/[0303]: acc-top1=51.132812 acc-top5=76.585938
Batch [0249]/[0303]: acc-top1=50.937500 acc-top5=76.756250
Batch [0299]/[0303]: acc-top1=50.609375 acc-top5=76.515625
[Epoch 014] training: accuracy=49.060588	 loss=2.175249
[Epoch 014] speed: 69 samples/sec	time cost: 3769.769041
[Epoch 014] validation: acc-top1=50.572401 acc-top5=76.469678 loss=2.194599
Epoch[015] Batch [0049]/[3760]	Speed: 47.776765 samples/sec	 accuracy=50.218750	 loss=2.167443	 lr=0.010000
Epoch[015] Batch [0099]/[3760]	Speed: 69.809745 samples/sec	 accuracy=50.203125	 loss=2.149319	 lr=0.010000
Epoch[015] Batch [0149]/[3760]	Speed: 69.247509 samples/sec	 accuracy=50.291667	 loss=2.137203	 lr=0.010000
Epoch[015] Batch [0199]/[3760]	Speed: 69.709556 samples/sec	 accuracy=50.257812	 loss=2.126334	 lr=0.010000
Epoch[015] Batch [0249]/[3760]	Speed: 69.924082 samples/sec	 accuracy=50.031250	 loss=2.132653	 lr=0.010000
Epoch[015] Batch [0299]/[3760]	Speed: 68.513296 samples/sec	 accuracy=50.057292	 loss=2.129696	 lr=0.010000
Epoch[015] Batch [0349]/[3760]	Speed: 70.710327 samples/sec	 accuracy=49.928571	 loss=2.132021	 lr=0.010000
Epoch[015] Batch [0399]/[3760]	Speed: 69.428848 samples/sec	 accuracy=49.812500	 loss=2.128080	 lr=0.010000
Epoch[015] Batch [0449]/[3760]	Speed: 70.305376 samples/sec	 accuracy=49.815972	 loss=2.129460	 lr=0.010000
Epoch[015] Batch [0499]/[3760]	Speed: 69.588909 samples/sec	 accuracy=49.715625	 loss=2.132524	 lr=0.010000
Epoch[015] Batch [0549]/[3760]	Speed: 68.532707 samples/sec	 accuracy=49.789773	 loss=2.131902	 lr=0.010000
Epoch[015] Batch [0599]/[3760]	Speed: 68.633454 samples/sec	 accuracy=49.890625	 loss=2.127631	 lr=0.010000
Epoch[015] Batch [0649]/[3760]	Speed: 70.966223 samples/sec	 accuracy=49.855769	 loss=2.125371	 lr=0.010000
Epoch[015] Batch [0699]/[3760]	Speed: 67.553304 samples/sec	 accuracy=49.957589	 loss=2.123204	 lr=0.010000
Epoch[015] Batch [0749]/[3760]	Speed: 72.469679 samples/sec	 accuracy=49.943750	 loss=2.123924	 lr=0.010000
Epoch[015] Batch [0799]/[3760]	Speed: 69.839865 samples/sec	 accuracy=50.025391	 loss=2.120802	 lr=0.010000
Epoch[015] Batch [0849]/[3760]	Speed: 68.593402 samples/sec	 accuracy=49.944853	 loss=2.125710	 lr=0.010000
Epoch[015] Batch [0899]/[3760]	Speed: 69.757908 samples/sec	 accuracy=49.947917	 loss=2.128960	 lr=0.010000
Epoch[015] Batch [0949]/[3760]	Speed: 70.002674 samples/sec	 accuracy=49.981908	 loss=2.128281	 lr=0.010000
Epoch[015] Batch [0999]/[3760]	Speed: 69.948115 samples/sec	 accuracy=49.956250	 loss=2.127200	 lr=0.010000
Epoch[015] Batch [1049]/[3760]	Speed: 68.975505 samples/sec	 accuracy=49.955357	 loss=2.130123	 lr=0.010000
Epoch[015] Batch [1099]/[3760]	Speed: 69.103162 samples/sec	 accuracy=49.842330	 loss=2.133633	 lr=0.010000
Epoch[015] Batch [1149]/[3760]	Speed: 69.386364 samples/sec	 accuracy=49.802989	 loss=2.136470	 lr=0.010000
Epoch[015] Batch [1199]/[3760]	Speed: 71.053794 samples/sec	 accuracy=49.863281	 loss=2.135147	 lr=0.010000
Epoch[015] Batch [1249]/[3760]	Speed: 67.203266 samples/sec	 accuracy=49.870000	 loss=2.136333	 lr=0.010000
Epoch[015] Batch [1299]/[3760]	Speed: 69.303152 samples/sec	 accuracy=49.853365	 loss=2.138516	 lr=0.010000
Epoch[015] Batch [1349]/[3760]	Speed: 70.137259 samples/sec	 accuracy=49.840278	 loss=2.139016	 lr=0.010000
Epoch[015] Batch [1399]/[3760]	Speed: 68.571066 samples/sec	 accuracy=49.786830	 loss=2.140536	 lr=0.010000
Epoch[015] Batch [1449]/[3760]	Speed: 68.737500 samples/sec	 accuracy=49.764009	 loss=2.142181	 lr=0.010000
Epoch[015] Batch [1499]/[3760]	Speed: 71.512148 samples/sec	 accuracy=49.746875	 loss=2.141080	 lr=0.010000
Epoch[015] Batch [1549]/[3760]	Speed: 68.927230 samples/sec	 accuracy=49.768145	 loss=2.140723	 lr=0.010000
Epoch[015] Batch [1599]/[3760]	Speed: 69.428435 samples/sec	 accuracy=49.755859	 loss=2.141075	 lr=0.010000
Epoch[015] Batch [1649]/[3760]	Speed: 70.298468 samples/sec	 accuracy=49.716856	 loss=2.142652	 lr=0.010000
Epoch[015] Batch [1699]/[3760]	Speed: 66.684688 samples/sec	 accuracy=49.693015	 loss=2.144276	 lr=0.010000
Epoch[015] Batch [1749]/[3760]	Speed: 72.058482 samples/sec	 accuracy=49.695536	 loss=2.143534	 lr=0.010000
Epoch[015] Batch [1799]/[3760]	Speed: 67.519642 samples/sec	 accuracy=49.723958	 loss=2.143049	 lr=0.010000
Epoch[015] Batch [1849]/[3760]	Speed: 69.993623 samples/sec	 accuracy=49.673142	 loss=2.145240	 lr=0.010000
Epoch[015] Batch [1899]/[3760]	Speed: 72.020594 samples/sec	 accuracy=49.649671	 loss=2.145388	 lr=0.010000
Epoch[015] Batch [1949]/[3760]	Speed: 69.937677 samples/sec	 accuracy=49.609776	 loss=2.146325	 lr=0.010000
Epoch[015] Batch [1999]/[3760]	Speed: 68.832774 samples/sec	 accuracy=49.599219	 loss=2.147446	 lr=0.010000
Epoch[015] Batch [2049]/[3760]	Speed: 69.131932 samples/sec	 accuracy=49.580793	 loss=2.147975	 lr=0.010000
Epoch[015] Batch [2099]/[3760]	Speed: 70.746773 samples/sec	 accuracy=49.607143	 loss=2.146710	 lr=0.010000
Epoch[015] Batch [2149]/[3760]	Speed: 67.669554 samples/sec	 accuracy=49.656977	 loss=2.144710	 lr=0.010000
Epoch[015] Batch [2199]/[3760]	Speed: 69.767457 samples/sec	 accuracy=49.706676	 loss=2.142775	 lr=0.010000
Epoch[015] Batch [2249]/[3760]	Speed: 68.627404 samples/sec	 accuracy=49.697222	 loss=2.143703	 lr=0.010000
Epoch[015] Batch [2299]/[3760]	Speed: 71.055397 samples/sec	 accuracy=49.721467	 loss=2.144373	 lr=0.010000
Epoch[015] Batch [2349]/[3760]	Speed: 70.463393 samples/sec	 accuracy=49.740027	 loss=2.143928	 lr=0.010000
Epoch[015] Batch [2399]/[3760]	Speed: 67.969136 samples/sec	 accuracy=49.753255	 loss=2.143667	 lr=0.010000
Epoch[015] Batch [2449]/[3760]	Speed: 68.513301 samples/sec	 accuracy=49.776148	 loss=2.142848	 lr=0.010000
Epoch[015] Batch [2499]/[3760]	Speed: 68.094456 samples/sec	 accuracy=49.745000	 loss=2.144581	 lr=0.010000
Epoch[015] Batch [2549]/[3760]	Speed: 71.079005 samples/sec	 accuracy=49.735294	 loss=2.144630	 lr=0.010000
Epoch[015] Batch [2599]/[3760]	Speed: 69.573736 samples/sec	 accuracy=49.730769	 loss=2.145477	 lr=0.010000
Epoch[015] Batch [2649]/[3760]	Speed: 70.305749 samples/sec	 accuracy=49.728184	 loss=2.145821	 lr=0.010000
Epoch[015] Batch [2699]/[3760]	Speed: 69.560541 samples/sec	 accuracy=49.717593	 loss=2.146548	 lr=0.010000
Epoch[015] Batch [2749]/[3760]	Speed: 69.767144 samples/sec	 accuracy=49.707386	 loss=2.146372	 lr=0.010000
Epoch[015] Batch [2799]/[3760]	Speed: 69.740649 samples/sec	 accuracy=49.691406	 loss=2.147001	 lr=0.010000
Epoch[015] Batch [2849]/[3760]	Speed: 68.092393 samples/sec	 accuracy=49.698465	 loss=2.146721	 lr=0.010000
Epoch[015] Batch [2899]/[3760]	Speed: 70.383017 samples/sec	 accuracy=49.720905	 loss=2.145979	 lr=0.010000
Epoch[015] Batch [2949]/[3760]	Speed: 68.318596 samples/sec	 accuracy=49.704979	 loss=2.146673	 lr=0.010000
Epoch[015] Batch [2999]/[3760]	Speed: 70.464841 samples/sec	 accuracy=49.708333	 loss=2.146999	 lr=0.010000
Epoch[015] Batch [3049]/[3760]	Speed: 69.445322 samples/sec	 accuracy=49.723873	 loss=2.146490	 lr=0.010000
Epoch[015] Batch [3099]/[3760]	Speed: 70.351250 samples/sec	 accuracy=49.746472	 loss=2.146146	 lr=0.010000
Epoch[015] Batch [3149]/[3760]	Speed: 68.935243 samples/sec	 accuracy=49.699405	 loss=2.147331	 lr=0.010000
Epoch[015] Batch [3199]/[3760]	Speed: 69.396841 samples/sec	 accuracy=49.716797	 loss=2.146733	 lr=0.010000
Epoch[015] Batch [3249]/[3760]	Speed: 70.268801 samples/sec	 accuracy=49.704808	 loss=2.146637	 lr=0.010000
Epoch[015] Batch [3299]/[3760]	Speed: 67.149238 samples/sec	 accuracy=49.690814	 loss=2.146866	 lr=0.010000
Epoch[015] Batch [3349]/[3760]	Speed: 71.319737 samples/sec	 accuracy=49.680504	 loss=2.147777	 lr=0.010000
Epoch[015] Batch [3399]/[3760]	Speed: 66.712444 samples/sec	 accuracy=49.676930	 loss=2.147813	 lr=0.010000
Epoch[015] Batch [3449]/[3760]	Speed: 71.958330 samples/sec	 accuracy=49.659420	 loss=2.148946	 lr=0.010000
Epoch[015] Batch [3499]/[3760]	Speed: 70.519458 samples/sec	 accuracy=49.639286	 loss=2.149682	 lr=0.010000
Epoch[015] Batch [3549]/[3760]	Speed: 69.294346 samples/sec	 accuracy=49.643486	 loss=2.149519	 lr=0.010000
Epoch[015] Batch [3599]/[3760]	Speed: 68.842986 samples/sec	 accuracy=49.647569	 loss=2.148769	 lr=0.010000
Epoch[015] Batch [3649]/[3760]	Speed: 69.647014 samples/sec	 accuracy=49.649401	 loss=2.147760	 lr=0.010000
Epoch[015] Batch [3699]/[3760]	Speed: 69.555383 samples/sec	 accuracy=49.638091	 loss=2.147874	 lr=0.010000
Epoch[015] Batch [3749]/[3760]	Speed: 77.681504 samples/sec	 accuracy=49.634167	 loss=2.148097	 lr=0.010000
Batch [0049]/[0303]: acc-top1=51.187500 acc-top5=76.906250
Batch [0099]/[0303]: acc-top1=51.187500 acc-top5=76.671875
Batch [0149]/[0303]: acc-top1=51.260417 acc-top5=77.062500
Batch [0199]/[0303]: acc-top1=51.625000 acc-top5=77.109375
Batch [0249]/[0303]: acc-top1=51.843750 acc-top5=77.137500
Batch [0299]/[0303]: acc-top1=51.604167 acc-top5=76.890625
[Epoch 015] training: accuracy=49.630984	 loss=2.148328
[Epoch 015] speed: 69 samples/sec	time cost: 3775.031940
[Epoch 015] validation: acc-top1=51.567657 acc-top5=76.882219 loss=2.169240
Epoch[016] Batch [0049]/[3760]	Speed: 48.133803 samples/sec	 accuracy=50.906250	 loss=2.108713	 lr=0.010000
Epoch[016] Batch [0099]/[3760]	Speed: 69.111038 samples/sec	 accuracy=50.578125	 loss=2.112152	 lr=0.010000
Epoch[016] Batch [0149]/[3760]	Speed: 67.435257 samples/sec	 accuracy=51.020833	 loss=2.090357	 lr=0.010000
Epoch[016] Batch [0199]/[3760]	Speed: 69.463035 samples/sec	 accuracy=50.851562	 loss=2.092385	 lr=0.010000
Epoch[016] Batch [0249]/[3760]	Speed: 70.328501 samples/sec	 accuracy=50.812500	 loss=2.096883	 lr=0.010000
Epoch[016] Batch [0299]/[3760]	Speed: 70.238410 samples/sec	 accuracy=50.468750	 loss=2.109928	 lr=0.010000
Epoch[016] Batch [0349]/[3760]	Speed: 68.551231 samples/sec	 accuracy=50.477679	 loss=2.107290	 lr=0.010000
Epoch[016] Batch [0399]/[3760]	Speed: 70.258271 samples/sec	 accuracy=50.675781	 loss=2.104412	 lr=0.010000
Epoch[016] Batch [0449]/[3760]	Speed: 68.205913 samples/sec	 accuracy=50.579861	 loss=2.106275	 lr=0.010000
Epoch[016] Batch [0499]/[3760]	Speed: 71.006998 samples/sec	 accuracy=50.521875	 loss=2.107593	 lr=0.010000
Epoch[016] Batch [0549]/[3760]	Speed: 70.070116 samples/sec	 accuracy=50.556818	 loss=2.106825	 lr=0.010000
Epoch[016] Batch [0599]/[3760]	Speed: 70.001754 samples/sec	 accuracy=50.635417	 loss=2.105936	 lr=0.010000
Epoch[016] Batch [0649]/[3760]	Speed: 68.990299 samples/sec	 accuracy=50.771635	 loss=2.099019	 lr=0.010000
Epoch[016] Batch [0699]/[3760]	Speed: 69.286277 samples/sec	 accuracy=50.705357	 loss=2.100637	 lr=0.010000
Epoch[016] Batch [0749]/[3760]	Speed: 68.218084 samples/sec	 accuracy=50.810417	 loss=2.101717	 lr=0.010000
Epoch[016] Batch [0799]/[3760]	Speed: 69.824595 samples/sec	 accuracy=50.818359	 loss=2.101639	 lr=0.010000
Epoch[016] Batch [0849]/[3760]	Speed: 70.675926 samples/sec	 accuracy=50.781250	 loss=2.101880	 lr=0.010000
Epoch[016] Batch [0899]/[3760]	Speed: 70.741201 samples/sec	 accuracy=50.763889	 loss=2.104822	 lr=0.010000
Epoch[016] Batch [0949]/[3760]	Speed: 69.638728 samples/sec	 accuracy=50.685855	 loss=2.107311	 lr=0.010000
Epoch[016] Batch [0999]/[3760]	Speed: 69.099523 samples/sec	 accuracy=50.709375	 loss=2.106038	 lr=0.010000
Epoch[016] Batch [1049]/[3760]	Speed: 69.561239 samples/sec	 accuracy=50.739583	 loss=2.104300	 lr=0.010000
Epoch[016] Batch [1099]/[3760]	Speed: 70.347854 samples/sec	 accuracy=50.717330	 loss=2.104766	 lr=0.010000
Epoch[016] Batch [1149]/[3760]	Speed: 68.124156 samples/sec	 accuracy=50.730978	 loss=2.103732	 lr=0.010000
Epoch[016] Batch [1199]/[3760]	Speed: 69.761286 samples/sec	 accuracy=50.701823	 loss=2.104068	 lr=0.010000
Epoch[016] Batch [1249]/[3760]	Speed: 70.250575 samples/sec	 accuracy=50.665000	 loss=2.105381	 lr=0.010000
Epoch[016] Batch [1299]/[3760]	Speed: 68.714762 samples/sec	 accuracy=50.610577	 loss=2.107310	 lr=0.010000
Epoch[016] Batch [1349]/[3760]	Speed: 69.643270 samples/sec	 accuracy=50.570602	 loss=2.108468	 lr=0.010000
Epoch[016] Batch [1399]/[3760]	Speed: 70.608752 samples/sec	 accuracy=50.617188	 loss=2.107194	 lr=0.010000
Epoch[016] Batch [1449]/[3760]	Speed: 70.391004 samples/sec	 accuracy=50.596983	 loss=2.107315	 lr=0.010000
Epoch[016] Batch [1499]/[3760]	Speed: 68.730937 samples/sec	 accuracy=50.581250	 loss=2.109617	 lr=0.010000
Epoch[016] Batch [1549]/[3760]	Speed: 69.641726 samples/sec	 accuracy=50.495968	 loss=2.113652	 lr=0.010000
Epoch[016] Batch [1599]/[3760]	Speed: 70.010432 samples/sec	 accuracy=50.518555	 loss=2.112891	 lr=0.010000
Epoch[016] Batch [1649]/[3760]	Speed: 71.112346 samples/sec	 accuracy=50.539773	 loss=2.111502	 lr=0.010000
Epoch[016] Batch [1699]/[3760]	Speed: 68.027648 samples/sec	 accuracy=50.544118	 loss=2.111996	 lr=0.010000
Epoch[016] Batch [1749]/[3760]	Speed: 70.638097 samples/sec	 accuracy=50.511607	 loss=2.111804	 lr=0.010000
Epoch[016] Batch [1799]/[3760]	Speed: 69.460601 samples/sec	 accuracy=50.506076	 loss=2.112659	 lr=0.010000
Epoch[016] Batch [1849]/[3760]	Speed: 68.562865 samples/sec	 accuracy=50.524493	 loss=2.111852	 lr=0.010000
Epoch[016] Batch [1899]/[3760]	Speed: 71.727106 samples/sec	 accuracy=50.504112	 loss=2.112438	 lr=0.010000
Epoch[016] Batch [1949]/[3760]	Speed: 67.145131 samples/sec	 accuracy=50.475160	 loss=2.113510	 lr=0.010000
Epoch[016] Batch [1999]/[3760]	Speed: 71.553884 samples/sec	 accuracy=50.446094	 loss=2.114065	 lr=0.010000
Epoch[016] Batch [2049]/[3760]	Speed: 70.307166 samples/sec	 accuracy=50.457317	 loss=2.113999	 lr=0.010000
Epoch[016] Batch [2099]/[3760]	Speed: 69.447189 samples/sec	 accuracy=50.410714	 loss=2.115780	 lr=0.010000
Epoch[016] Batch [2149]/[3760]	Speed: 67.198985 samples/sec	 accuracy=50.413517	 loss=2.116259	 lr=0.010000
Epoch[016] Batch [2199]/[3760]	Speed: 71.540191 samples/sec	 accuracy=50.400568	 loss=2.117338	 lr=0.010000
Epoch[016] Batch [2249]/[3760]	Speed: 69.782494 samples/sec	 accuracy=50.372222	 loss=2.118993	 lr=0.010000
Epoch[016] Batch [2299]/[3760]	Speed: 69.339024 samples/sec	 accuracy=50.336957	 loss=2.120603	 lr=0.010000
Epoch[016] Batch [2349]/[3760]	Speed: 70.234614 samples/sec	 accuracy=50.319149	 loss=2.120801	 lr=0.010000
Epoch[016] Batch [2399]/[3760]	Speed: 68.021367 samples/sec	 accuracy=50.305990	 loss=2.121326	 lr=0.010000
Epoch[016] Batch [2449]/[3760]	Speed: 70.379156 samples/sec	 accuracy=50.314413	 loss=2.121087	 lr=0.010000
Epoch[016] Batch [2499]/[3760]	Speed: 70.206037 samples/sec	 accuracy=50.312500	 loss=2.121615	 lr=0.010000
Epoch[016] Batch [2549]/[3760]	Speed: 70.391697 samples/sec	 accuracy=50.299020	 loss=2.122044	 lr=0.010000
Epoch[016] Batch [2599]/[3760]	Speed: 67.533768 samples/sec	 accuracy=50.274639	 loss=2.122726	 lr=0.010000
Epoch[016] Batch [2649]/[3760]	Speed: 70.686902 samples/sec	 accuracy=50.280660	 loss=2.122727	 lr=0.010000
Epoch[016] Batch [2699]/[3760]	Speed: 68.949889 samples/sec	 accuracy=50.284722	 loss=2.123054	 lr=0.010000
Epoch[016] Batch [2749]/[3760]	Speed: 69.141505 samples/sec	 accuracy=50.264773	 loss=2.123486	 lr=0.010000
Epoch[016] Batch [2799]/[3760]	Speed: 70.652259 samples/sec	 accuracy=50.223772	 loss=2.124685	 lr=0.010000
Epoch[016] Batch [2849]/[3760]	Speed: 65.332821 samples/sec	 accuracy=50.237939	 loss=2.123811	 lr=0.010000
Epoch[016] Batch [2899]/[3760]	Speed: 72.675777 samples/sec	 accuracy=50.215517	 loss=2.124592	 lr=0.010000
Epoch[016] Batch [2949]/[3760]	Speed: 69.255325 samples/sec	 accuracy=50.211335	 loss=2.124600	 lr=0.010000
Epoch[016] Batch [2999]/[3760]	Speed: 70.604621 samples/sec	 accuracy=50.225521	 loss=2.124288	 lr=0.010000
Epoch[016] Batch [3049]/[3760]	Speed: 69.972842 samples/sec	 accuracy=50.189037	 loss=2.125311	 lr=0.010000
Epoch[016] Batch [3099]/[3760]	Speed: 69.399670 samples/sec	 accuracy=50.168851	 loss=2.125842	 lr=0.010000
Epoch[016] Batch [3149]/[3760]	Speed: 70.275683 samples/sec	 accuracy=50.169643	 loss=2.125983	 lr=0.010000
Epoch[016] Batch [3199]/[3760]	Speed: 68.540961 samples/sec	 accuracy=50.166992	 loss=2.125706	 lr=0.010000
Epoch[016] Batch [3249]/[3760]	Speed: 68.800565 samples/sec	 accuracy=50.156731	 loss=2.125848	 lr=0.010000
Epoch[016] Batch [3299]/[3760]	Speed: 70.688178 samples/sec	 accuracy=50.137784	 loss=2.126523	 lr=0.010000
Epoch[016] Batch [3349]/[3760]	Speed: 69.622905 samples/sec	 accuracy=50.143190	 loss=2.125932	 lr=0.010000
Epoch[016] Batch [3399]/[3760]	Speed: 70.652250 samples/sec	 accuracy=50.148897	 loss=2.125884	 lr=0.010000
Epoch[016] Batch [3449]/[3760]	Speed: 67.753934 samples/sec	 accuracy=50.135417	 loss=2.125814	 lr=0.010000
Epoch[016] Batch [3499]/[3760]	Speed: 70.932331 samples/sec	 accuracy=50.127232	 loss=2.126018	 lr=0.010000
Epoch[016] Batch [3549]/[3760]	Speed: 69.646608 samples/sec	 accuracy=50.113116	 loss=2.126790	 lr=0.010000
Epoch[016] Batch [3599]/[3760]	Speed: 67.798812 samples/sec	 accuracy=50.109809	 loss=2.126755	 lr=0.010000
Epoch[016] Batch [3649]/[3760]	Speed: 70.771557 samples/sec	 accuracy=50.102740	 loss=2.127329	 lr=0.010000
Epoch[016] Batch [3699]/[3760]	Speed: 69.716627 samples/sec	 accuracy=50.113176	 loss=2.126939	 lr=0.010000
Epoch[016] Batch [3749]/[3760]	Speed: 76.357920 samples/sec	 accuracy=50.116667	 loss=2.127132	 lr=0.010000
Batch [0049]/[0303]: acc-top1=50.062500 acc-top5=76.500000
Batch [0099]/[0303]: acc-top1=49.906250 acc-top5=76.125000
Batch [0149]/[0303]: acc-top1=50.541667 acc-top5=76.437500
Batch [0199]/[0303]: acc-top1=50.929688 acc-top5=76.695312
Batch [0249]/[0303]: acc-top1=50.987500 acc-top5=76.793750
Batch [0299]/[0303]: acc-top1=50.989583 acc-top5=76.562500
[Epoch 016] training: accuracy=50.117603	 loss=2.127177
[Epoch 016] speed: 69 samples/sec	time cost: 3766.490277
[Epoch 016] validation: acc-top1=50.948845 acc-top5=76.500619 loss=2.210744
Epoch[017] Batch [0049]/[3759]	Speed: 47.681343 samples/sec	 accuracy=50.625000	 loss=2.075466	 lr=0.010000
Epoch[017] Batch [0099]/[3759]	Speed: 68.505191 samples/sec	 accuracy=51.906250	 loss=2.046665	 lr=0.010000
Epoch[017] Batch [0149]/[3759]	Speed: 69.372869 samples/sec	 accuracy=51.343750	 loss=2.065342	 lr=0.010000
Epoch[017] Batch [0199]/[3759]	Speed: 70.208128 samples/sec	 accuracy=51.460938	 loss=2.066220	 lr=0.010000
Epoch[017] Batch [0249]/[3759]	Speed: 67.498915 samples/sec	 accuracy=51.318750	 loss=2.071162	 lr=0.010000
Epoch[017] Batch [0299]/[3759]	Speed: 71.768477 samples/sec	 accuracy=51.213542	 loss=2.065642	 lr=0.010000
Epoch[017] Batch [0349]/[3759]	Speed: 68.425884 samples/sec	 accuracy=51.200893	 loss=2.069646	 lr=0.010000
Epoch[017] Batch [0399]/[3759]	Speed: 70.131906 samples/sec	 accuracy=51.152344	 loss=2.073081	 lr=0.010000
Epoch[017] Batch [0449]/[3759]	Speed: 68.481849 samples/sec	 accuracy=51.194444	 loss=2.072326	 lr=0.010000
Epoch[017] Batch [0499]/[3759]	Speed: 70.160916 samples/sec	 accuracy=51.100000	 loss=2.072169	 lr=0.010000
Epoch[017] Batch [0549]/[3759]	Speed: 68.022172 samples/sec	 accuracy=51.136364	 loss=2.075934	 lr=0.010000
Epoch[017] Batch [0599]/[3759]	Speed: 70.457957 samples/sec	 accuracy=51.171875	 loss=2.076385	 lr=0.010000
Epoch[017] Batch [0649]/[3759]	Speed: 70.617275 samples/sec	 accuracy=51.103365	 loss=2.076367	 lr=0.010000
Epoch[017] Batch [0699]/[3759]	Speed: 70.099146 samples/sec	 accuracy=51.087054	 loss=2.079833	 lr=0.010000
Epoch[017] Batch [0749]/[3759]	Speed: 69.447464 samples/sec	 accuracy=51.097917	 loss=2.078886	 lr=0.010000
Epoch[017] Batch [0799]/[3759]	Speed: 70.147845 samples/sec	 accuracy=51.011719	 loss=2.080721	 lr=0.010000
Epoch[017] Batch [0849]/[3759]	Speed: 69.450619 samples/sec	 accuracy=50.911765	 loss=2.087408	 lr=0.010000
Epoch[017] Batch [0899]/[3759]	Speed: 69.127666 samples/sec	 accuracy=50.850694	 loss=2.091353	 lr=0.010000
Epoch[017] Batch [0949]/[3759]	Speed: 71.344823 samples/sec	 accuracy=50.827303	 loss=2.091720	 lr=0.010000
Epoch[017] Batch [0999]/[3759]	Speed: 70.278610 samples/sec	 accuracy=50.778125	 loss=2.093086	 lr=0.010000
Epoch[017] Batch [1049]/[3759]	Speed: 67.312925 samples/sec	 accuracy=50.747024	 loss=2.093867	 lr=0.010000
Epoch[017] Batch [1099]/[3759]	Speed: 69.739153 samples/sec	 accuracy=50.717330	 loss=2.097145	 lr=0.010000
Epoch[017] Batch [1149]/[3759]	Speed: 68.930899 samples/sec	 accuracy=50.706522	 loss=2.096875	 lr=0.010000
Epoch[017] Batch [1199]/[3759]	Speed: 69.899235 samples/sec	 accuracy=50.714844	 loss=2.095295	 lr=0.010000
Epoch[017] Batch [1249]/[3759]	Speed: 70.725214 samples/sec	 accuracy=50.740000	 loss=2.093339	 lr=0.010000
Epoch[017] Batch [1299]/[3759]	Speed: 69.450434 samples/sec	 accuracy=50.710337	 loss=2.095232	 lr=0.010000
Epoch[017] Batch [1349]/[3759]	Speed: 68.843200 samples/sec	 accuracy=50.655093	 loss=2.097498	 lr=0.010000
Epoch[017] Batch [1399]/[3759]	Speed: 69.314050 samples/sec	 accuracy=50.679688	 loss=2.097467	 lr=0.010000
Epoch[017] Batch [1449]/[3759]	Speed: 69.983198 samples/sec	 accuracy=50.645474	 loss=2.097799	 lr=0.010000
Epoch[017] Batch [1499]/[3759]	Speed: 69.755641 samples/sec	 accuracy=50.636458	 loss=2.098710	 lr=0.010000
Epoch[017] Batch [1549]/[3759]	Speed: 68.788672 samples/sec	 accuracy=50.610887	 loss=2.100908	 lr=0.010000
Epoch[017] Batch [1599]/[3759]	Speed: 70.972606 samples/sec	 accuracy=50.616211	 loss=2.100552	 lr=0.010000
Epoch[017] Batch [1649]/[3759]	Speed: 69.016243 samples/sec	 accuracy=50.651515	 loss=2.101404	 lr=0.010000
Epoch[017] Batch [1699]/[3759]	Speed: 68.604082 samples/sec	 accuracy=50.634191	 loss=2.101891	 lr=0.010000
Epoch[017] Batch [1749]/[3759]	Speed: 70.752390 samples/sec	 accuracy=50.643750	 loss=2.101253	 lr=0.010000
Epoch[017] Batch [1799]/[3759]	Speed: 69.311992 samples/sec	 accuracy=50.645833	 loss=2.100695	 lr=0.010000
Epoch[017] Batch [1849]/[3759]	Speed: 68.214105 samples/sec	 accuracy=50.655405	 loss=2.100406	 lr=0.010000
Epoch[017] Batch [1899]/[3759]	Speed: 72.135134 samples/sec	 accuracy=50.666941	 loss=2.100738	 lr=0.010000
Epoch[017] Batch [1949]/[3759]	Speed: 70.106744 samples/sec	 accuracy=50.663462	 loss=2.101567	 lr=0.010000
Epoch[017] Batch [1999]/[3759]	Speed: 69.722303 samples/sec	 accuracy=50.672656	 loss=2.102780	 lr=0.010000
Epoch[017] Batch [2049]/[3759]	Speed: 69.540395 samples/sec	 accuracy=50.685976	 loss=2.101600	 lr=0.010000
Epoch[017] Batch [2099]/[3759]	Speed: 69.619523 samples/sec	 accuracy=50.674107	 loss=2.102349	 lr=0.010000
Epoch[017] Batch [2149]/[3759]	Speed: 69.103687 samples/sec	 accuracy=50.689680	 loss=2.101611	 lr=0.010000
Epoch[017] Batch [2199]/[3759]	Speed: 71.012374 samples/sec	 accuracy=50.691761	 loss=2.101942	 lr=0.010000
Epoch[017] Batch [2249]/[3759]	Speed: 69.463098 samples/sec	 accuracy=50.680556	 loss=2.102126	 lr=0.010000
Epoch[017] Batch [2299]/[3759]	Speed: 69.801796 samples/sec	 accuracy=50.641984	 loss=2.102986	 lr=0.010000
Epoch[017] Batch [2349]/[3759]	Speed: 70.227880 samples/sec	 accuracy=50.646277	 loss=2.103617	 lr=0.010000
Epoch[017] Batch [2399]/[3759]	Speed: 68.173794 samples/sec	 accuracy=50.652344	 loss=2.103428	 lr=0.010000
Epoch[017] Batch [2449]/[3759]	Speed: 70.266063 samples/sec	 accuracy=50.631378	 loss=2.104338	 lr=0.010000
Epoch[017] Batch [2499]/[3759]	Speed: 70.427677 samples/sec	 accuracy=50.630625	 loss=2.105218	 lr=0.010000
Epoch[017] Batch [2549]/[3759]	Speed: 68.558698 samples/sec	 accuracy=50.623162	 loss=2.105428	 lr=0.010000
Epoch[017] Batch [2599]/[3759]	Speed: 68.793998 samples/sec	 accuracy=50.620793	 loss=2.104668	 lr=0.010000
Epoch[017] Batch [2649]/[3759]	Speed: 70.743675 samples/sec	 accuracy=50.604363	 loss=2.105451	 lr=0.010000
Epoch[017] Batch [2699]/[3759]	Speed: 70.360945 samples/sec	 accuracy=50.608796	 loss=2.104738	 lr=0.010000
Epoch[017] Batch [2749]/[3759]	Speed: 68.829547 samples/sec	 accuracy=50.634091	 loss=2.103454	 lr=0.010000
Epoch[017] Batch [2799]/[3759]	Speed: 69.636483 samples/sec	 accuracy=50.642299	 loss=2.103724	 lr=0.010000
Epoch[017] Batch [2849]/[3759]	Speed: 69.960035 samples/sec	 accuracy=50.630482	 loss=2.104006	 lr=0.010000
Epoch[017] Batch [2899]/[3759]	Speed: 70.173065 samples/sec	 accuracy=50.619612	 loss=2.104763	 lr=0.010000
Epoch[017] Batch [2949]/[3759]	Speed: 68.326676 samples/sec	 accuracy=50.628178	 loss=2.105506	 lr=0.010000
Epoch[017] Batch [2999]/[3759]	Speed: 69.863369 samples/sec	 accuracy=50.652083	 loss=2.104833	 lr=0.010000
Epoch[017] Batch [3049]/[3759]	Speed: 69.230869 samples/sec	 accuracy=50.660861	 loss=2.104257	 lr=0.010000
Epoch[017] Batch [3099]/[3759]	Speed: 70.245409 samples/sec	 accuracy=50.659274	 loss=2.104646	 lr=0.010000
Epoch[017] Batch [3149]/[3759]	Speed: 69.875859 samples/sec	 accuracy=50.641369	 loss=2.105085	 lr=0.010000
Epoch[017] Batch [3199]/[3759]	Speed: 69.217409 samples/sec	 accuracy=50.621094	 loss=2.105358	 lr=0.010000
Epoch[017] Batch [3249]/[3759]	Speed: 70.077476 samples/sec	 accuracy=50.592788	 loss=2.106010	 lr=0.010000
Epoch[017] Batch [3299]/[3759]	Speed: 69.499812 samples/sec	 accuracy=50.580019	 loss=2.106472	 lr=0.010000
Epoch[017] Batch [3349]/[3759]	Speed: 69.675002 samples/sec	 accuracy=50.572761	 loss=2.106688	 lr=0.010000
Epoch[017] Batch [3399]/[3759]	Speed: 70.375869 samples/sec	 accuracy=50.569393	 loss=2.106423	 lr=0.010000
Epoch[017] Batch [3449]/[3759]	Speed: 69.292473 samples/sec	 accuracy=50.548007	 loss=2.107563	 lr=0.010000
Epoch[017] Batch [3499]/[3759]	Speed: 69.639697 samples/sec	 accuracy=50.551786	 loss=2.107476	 lr=0.010000
Epoch[017] Batch [3549]/[3759]	Speed: 68.837704 samples/sec	 accuracy=50.548415	 loss=2.107446	 lr=0.010000
Epoch[017] Batch [3599]/[3759]	Speed: 70.360518 samples/sec	 accuracy=50.544271	 loss=2.107794	 lr=0.010000
Epoch[017] Batch [3649]/[3759]	Speed: 69.054812 samples/sec	 accuracy=50.514983	 loss=2.108553	 lr=0.010000
Epoch[017] Batch [3699]/[3759]	Speed: 69.953300 samples/sec	 accuracy=50.526182	 loss=2.108096	 lr=0.010000
Epoch[017] Batch [3749]/[3759]	Speed: 76.989403 samples/sec	 accuracy=50.509583	 loss=2.108785	 lr=0.010000
Batch [0049]/[0303]: acc-top1=51.062500 acc-top5=76.406250
Batch [0099]/[0303]: acc-top1=50.984375 acc-top5=76.437500
Batch [0149]/[0303]: acc-top1=51.572917 acc-top5=76.812500
Batch [0199]/[0303]: acc-top1=51.875000 acc-top5=76.937500
Batch [0249]/[0303]: acc-top1=52.181250 acc-top5=77.075000
Batch [0299]/[0303]: acc-top1=52.020833 acc-top5=76.901042
[Epoch 017] training: accuracy=50.500050	 loss=2.109088
[Epoch 017] speed: 69 samples/sec	time cost: 3765.035787
[Epoch 017] validation: acc-top1=52.005982 acc-top5=76.856436 loss=2.130524
Epoch[018] Batch [0049]/[3760]	Speed: 49.137903 samples/sec	 accuracy=50.000000	 loss=2.070467	 lr=0.010000
Epoch[018] Batch [0099]/[3760]	Speed: 67.993350 samples/sec	 accuracy=50.734375	 loss=2.064803	 lr=0.010000
Epoch[018] Batch [0149]/[3760]	Speed: 70.509150 samples/sec	 accuracy=50.822917	 loss=2.064663	 lr=0.010000
Epoch[018] Batch [0199]/[3760]	Speed: 68.641824 samples/sec	 accuracy=51.203125	 loss=2.062630	 lr=0.010000
Epoch[018] Batch [0249]/[3760]	Speed: 68.652067 samples/sec	 accuracy=51.250000	 loss=2.062412	 lr=0.010000
Epoch[018] Batch [0299]/[3760]	Speed: 70.258201 samples/sec	 accuracy=51.234375	 loss=2.069135	 lr=0.010000
Epoch[018] Batch [0349]/[3760]	Speed: 70.794263 samples/sec	 accuracy=51.383929	 loss=2.068517	 lr=0.010000
Epoch[018] Batch [0399]/[3760]	Speed: 70.120604 samples/sec	 accuracy=51.414062	 loss=2.070719	 lr=0.010000
Epoch[018] Batch [0449]/[3760]	Speed: 70.093537 samples/sec	 accuracy=51.277778	 loss=2.078099	 lr=0.010000
Epoch[018] Batch [0499]/[3760]	Speed: 68.810219 samples/sec	 accuracy=51.309375	 loss=2.080495	 lr=0.010000
Epoch[018] Batch [0549]/[3760]	Speed: 70.480542 samples/sec	 accuracy=51.218750	 loss=2.078963	 lr=0.010000
Epoch[018] Batch [0599]/[3760]	Speed: 70.260393 samples/sec	 accuracy=51.322917	 loss=2.075228	 lr=0.010000
Epoch[018] Batch [0649]/[3760]	Speed: 65.738506 samples/sec	 accuracy=51.283654	 loss=2.077285	 lr=0.010000
Epoch[018] Batch [0699]/[3760]	Speed: 71.369823 samples/sec	 accuracy=51.203125	 loss=2.075984	 lr=0.010000
Epoch[018] Batch [0749]/[3760]	Speed: 70.401151 samples/sec	 accuracy=51.233333	 loss=2.074077	 lr=0.010000
Epoch[018] Batch [0799]/[3760]	Speed: 69.449187 samples/sec	 accuracy=51.228516	 loss=2.072710	 lr=0.010000
Epoch[018] Batch [0849]/[3760]	Speed: 70.097348 samples/sec	 accuracy=51.215074	 loss=2.072518	 lr=0.010000
Epoch[018] Batch [0899]/[3760]	Speed: 69.487877 samples/sec	 accuracy=51.288194	 loss=2.070087	 lr=0.010000
Epoch[018] Batch [0949]/[3760]	Speed: 69.771369 samples/sec	 accuracy=51.325658	 loss=2.071962	 lr=0.010000
Epoch[018] Batch [0999]/[3760]	Speed: 68.354946 samples/sec	 accuracy=51.284375	 loss=2.074043	 lr=0.010000
Epoch[018] Batch [1049]/[3760]	Speed: 67.497535 samples/sec	 accuracy=51.215774	 loss=2.075150	 lr=0.010000
Epoch[018] Batch [1099]/[3760]	Speed: 72.146093 samples/sec	 accuracy=51.244318	 loss=2.076792	 lr=0.010000
Epoch[018] Batch [1149]/[3760]	Speed: 68.660025 samples/sec	 accuracy=51.218750	 loss=2.076710	 lr=0.010000
Epoch[018] Batch [1199]/[3760]	Speed: 71.010763 samples/sec	 accuracy=51.177083	 loss=2.077768	 lr=0.010000
Epoch[018] Batch [1249]/[3760]	Speed: 67.898298 samples/sec	 accuracy=51.118750	 loss=2.077950	 lr=0.010000
Epoch[018] Batch [1299]/[3760]	Speed: 69.804264 samples/sec	 accuracy=51.144231	 loss=2.078454	 lr=0.010000
Epoch[018] Batch [1349]/[3760]	Speed: 69.770241 samples/sec	 accuracy=51.116898	 loss=2.079118	 lr=0.010000
Epoch[018] Batch [1399]/[3760]	Speed: 68.561410 samples/sec	 accuracy=51.123884	 loss=2.077260	 lr=0.010000
Epoch[018] Batch [1449]/[3760]	Speed: 67.461021 samples/sec	 accuracy=51.119612	 loss=2.078613	 lr=0.010000
Epoch[018] Batch [1499]/[3760]	Speed: 71.348584 samples/sec	 accuracy=51.154167	 loss=2.077373	 lr=0.010000
Epoch[018] Batch [1549]/[3760]	Speed: 70.008044 samples/sec	 accuracy=51.194556	 loss=2.075546	 lr=0.010000
Epoch[018] Batch [1599]/[3760]	Speed: 70.268100 samples/sec	 accuracy=51.154297	 loss=2.077635	 lr=0.010000
Epoch[018] Batch [1649]/[3760]	Speed: 69.305742 samples/sec	 accuracy=51.145833	 loss=2.079085	 lr=0.010000
Epoch[018] Batch [1699]/[3760]	Speed: 69.935494 samples/sec	 accuracy=51.142463	 loss=2.079596	 lr=0.010000
Epoch[018] Batch [1749]/[3760]	Speed: 68.344858 samples/sec	 accuracy=51.077679	 loss=2.081386	 lr=0.010000
Epoch[018] Batch [1799]/[3760]	Speed: 70.961591 samples/sec	 accuracy=51.091146	 loss=2.080383	 lr=0.010000
Epoch[018] Batch [1849]/[3760]	Speed: 68.764836 samples/sec	 accuracy=51.113176	 loss=2.080535	 lr=0.010000
Epoch[018] Batch [1899]/[3760]	Speed: 69.371822 samples/sec	 accuracy=51.115132	 loss=2.080582	 lr=0.010000
Epoch[018] Batch [1949]/[3760]	Speed: 68.206970 samples/sec	 accuracy=51.086538	 loss=2.081740	 lr=0.010000
Epoch[018] Batch [1999]/[3760]	Speed: 69.718020 samples/sec	 accuracy=51.067188	 loss=2.083116	 lr=0.010000
Epoch[018] Batch [2049]/[3760]	Speed: 67.164245 samples/sec	 accuracy=51.025915	 loss=2.084848	 lr=0.010000
Epoch[018] Batch [2099]/[3760]	Speed: 72.486951 samples/sec	 accuracy=51.031994	 loss=2.084224	 lr=0.010000
Epoch[018] Batch [2149]/[3760]	Speed: 69.574220 samples/sec	 accuracy=51.005814	 loss=2.085281	 lr=0.010000
Epoch[018] Batch [2199]/[3760]	Speed: 66.066409 samples/sec	 accuracy=51.002841	 loss=2.086040	 lr=0.010000
Epoch[018] Batch [2249]/[3760]	Speed: 71.238501 samples/sec	 accuracy=50.977778	 loss=2.087491	 lr=0.010000
Epoch[018] Batch [2299]/[3760]	Speed: 65.722530 samples/sec	 accuracy=50.964674	 loss=2.086965	 lr=0.010000
Epoch[018] Batch [2349]/[3760]	Speed: 73.326161 samples/sec	 accuracy=50.942154	 loss=2.087671	 lr=0.010000
Epoch[018] Batch [2399]/[3760]	Speed: 70.815754 samples/sec	 accuracy=50.933594	 loss=2.087642	 lr=0.010000
Epoch[018] Batch [2449]/[3760]	Speed: 68.263693 samples/sec	 accuracy=50.902423	 loss=2.089056	 lr=0.010000
Epoch[018] Batch [2499]/[3760]	Speed: 70.137710 samples/sec	 accuracy=50.893750	 loss=2.090382	 lr=0.010000
Epoch[018] Batch [2549]/[3760]	Speed: 70.251911 samples/sec	 accuracy=50.922181	 loss=2.089654	 lr=0.010000
Epoch[018] Batch [2599]/[3760]	Speed: 67.512765 samples/sec	 accuracy=50.890625	 loss=2.090380	 lr=0.010000
Epoch[018] Batch [2649]/[3760]	Speed: 70.553860 samples/sec	 accuracy=50.877948	 loss=2.091165	 lr=0.010000
Epoch[018] Batch [2699]/[3760]	Speed: 70.063553 samples/sec	 accuracy=50.851852	 loss=2.093421	 lr=0.010000
Epoch[018] Batch [2749]/[3760]	Speed: 69.981399 samples/sec	 accuracy=50.872159	 loss=2.092331	 lr=0.010000
Epoch[018] Batch [2799]/[3760]	Speed: 70.371593 samples/sec	 accuracy=50.866071	 loss=2.091463	 lr=0.010000
Epoch[018] Batch [2849]/[3760]	Speed: 69.787314 samples/sec	 accuracy=50.854715	 loss=2.092156	 lr=0.010000
Epoch[018] Batch [2899]/[3760]	Speed: 69.115602 samples/sec	 accuracy=50.856142	 loss=2.091710	 lr=0.010000
Epoch[018] Batch [2949]/[3760]	Speed: 69.589660 samples/sec	 accuracy=50.872881	 loss=2.091532	 lr=0.010000
Epoch[018] Batch [2999]/[3760]	Speed: 70.893712 samples/sec	 accuracy=50.848437	 loss=2.092181	 lr=0.010000
Epoch[018] Batch [3049]/[3760]	Speed: 69.166423 samples/sec	 accuracy=50.830943	 loss=2.092906	 lr=0.010000
Epoch[018] Batch [3099]/[3760]	Speed: 70.036062 samples/sec	 accuracy=50.831653	 loss=2.092155	 lr=0.010000
Epoch[018] Batch [3149]/[3760]	Speed: 68.806138 samples/sec	 accuracy=50.836310	 loss=2.091894	 lr=0.010000
Epoch[018] Batch [3199]/[3760]	Speed: 70.042562 samples/sec	 accuracy=50.824219	 loss=2.092239	 lr=0.010000
Epoch[018] Batch [3249]/[3760]	Speed: 70.763746 samples/sec	 accuracy=50.826442	 loss=2.092019	 lr=0.010000
Epoch[018] Batch [3299]/[3760]	Speed: 69.229862 samples/sec	 accuracy=50.806818	 loss=2.092461	 lr=0.010000
Epoch[018] Batch [3349]/[3760]	Speed: 69.849331 samples/sec	 accuracy=50.810168	 loss=2.092237	 lr=0.010000
Epoch[018] Batch [3399]/[3760]	Speed: 68.494916 samples/sec	 accuracy=50.815257	 loss=2.092261	 lr=0.010000
Epoch[018] Batch [3449]/[3760]	Speed: 69.950116 samples/sec	 accuracy=50.826540	 loss=2.092041	 lr=0.010000
Epoch[018] Batch [3499]/[3760]	Speed: 69.701430 samples/sec	 accuracy=50.822768	 loss=2.092558	 lr=0.010000
Epoch[018] Batch [3549]/[3760]	Speed: 70.366658 samples/sec	 accuracy=50.818662	 loss=2.093265	 lr=0.010000
Epoch[018] Batch [3599]/[3760]	Speed: 69.257430 samples/sec	 accuracy=50.809896	 loss=2.093689	 lr=0.010000
Epoch[018] Batch [3649]/[3760]	Speed: 70.011728 samples/sec	 accuracy=50.790240	 loss=2.094795	 lr=0.010000
Epoch[018] Batch [3699]/[3760]	Speed: 68.966269 samples/sec	 accuracy=50.786740	 loss=2.094880	 lr=0.010000
Epoch[018] Batch [3749]/[3760]	Speed: 77.773209 samples/sec	 accuracy=50.777083	 loss=2.095026	 lr=0.010000
Batch [0049]/[0303]: acc-top1=52.187500 acc-top5=76.218750
Batch [0099]/[0303]: acc-top1=52.296875 acc-top5=76.359375
Batch [0149]/[0303]: acc-top1=52.510417 acc-top5=77.031250
Batch [0199]/[0303]: acc-top1=52.929688 acc-top5=77.460938
Batch [0249]/[0303]: acc-top1=53.056250 acc-top5=77.650000
Batch [0299]/[0303]: acc-top1=52.677083 acc-top5=77.312500
[Epoch 018] training: accuracy=50.774601	 loss=2.094920
[Epoch 018] speed: 69 samples/sec	time cost: 3765.673129
[Epoch 018] validation: acc-top1=52.676361 acc-top5=77.294761 loss=2.166220
Epoch[019] Batch [0049]/[3760]	Speed: 49.054227 samples/sec	 accuracy=50.937500	 loss=2.076410	 lr=0.010000
Epoch[019] Batch [0099]/[3760]	Speed: 67.993719 samples/sec	 accuracy=52.421875	 loss=2.016517	 lr=0.010000
Epoch[019] Batch [0149]/[3760]	Speed: 67.492178 samples/sec	 accuracy=52.187500	 loss=2.025256	 lr=0.010000
Epoch[019] Batch [0199]/[3760]	Speed: 68.002306 samples/sec	 accuracy=52.296875	 loss=2.020847	 lr=0.010000
Epoch[019] Batch [0249]/[3760]	Speed: 72.232118 samples/sec	 accuracy=51.993750	 loss=2.022986	 lr=0.010000
Epoch[019] Batch [0299]/[3760]	Speed: 69.865054 samples/sec	 accuracy=51.963542	 loss=2.026002	 lr=0.010000
Epoch[019] Batch [0349]/[3760]	Speed: 70.234216 samples/sec	 accuracy=52.071429	 loss=2.022585	 lr=0.010000
Epoch[019] Batch [0399]/[3760]	Speed: 70.916602 samples/sec	 accuracy=52.089844	 loss=2.026105	 lr=0.010000
Epoch[019] Batch [0449]/[3760]	Speed: 67.035647 samples/sec	 accuracy=52.055556	 loss=2.030388	 lr=0.010000
Epoch[019] Batch [0499]/[3760]	Speed: 71.787540 samples/sec	 accuracy=52.028125	 loss=2.037093	 lr=0.010000
Epoch[019] Batch [0549]/[3760]	Speed: 70.472632 samples/sec	 accuracy=52.065341	 loss=2.037053	 lr=0.010000
Epoch[019] Batch [0599]/[3760]	Speed: 69.032265 samples/sec	 accuracy=51.833333	 loss=2.047064	 lr=0.010000
Epoch[019] Batch [0649]/[3760]	Speed: 70.258851 samples/sec	 accuracy=51.790865	 loss=2.047664	 lr=0.010000
Epoch[019] Batch [0699]/[3760]	Speed: 68.674785 samples/sec	 accuracy=51.792411	 loss=2.045075	 lr=0.010000
Epoch[019] Batch [0749]/[3760]	Speed: 71.227118 samples/sec	 accuracy=51.885417	 loss=2.043566	 lr=0.010000
Epoch[019] Batch [0799]/[3760]	Speed: 70.304389 samples/sec	 accuracy=51.830078	 loss=2.045135	 lr=0.010000
Epoch[019] Batch [0849]/[3760]	Speed: 68.666048 samples/sec	 accuracy=51.766544	 loss=2.049952	 lr=0.010000
Epoch[019] Batch [0899]/[3760]	Speed: 68.777525 samples/sec	 accuracy=51.670139	 loss=2.051129	 lr=0.010000
Epoch[019] Batch [0949]/[3760]	Speed: 70.025741 samples/sec	 accuracy=51.623355	 loss=2.052195	 lr=0.010000
Epoch[019] Batch [0999]/[3760]	Speed: 70.756316 samples/sec	 accuracy=51.612500	 loss=2.050340	 lr=0.010000
Epoch[019] Batch [1049]/[3760]	Speed: 70.095295 samples/sec	 accuracy=51.581845	 loss=2.051812	 lr=0.010000
Epoch[019] Batch [1099]/[3760]	Speed: 70.550093 samples/sec	 accuracy=51.661932	 loss=2.048965	 lr=0.010000
Epoch[019] Batch [1149]/[3760]	Speed: 68.330952 samples/sec	 accuracy=51.673913	 loss=2.049012	 lr=0.010000
Epoch[019] Batch [1199]/[3760]	Speed: 69.045196 samples/sec	 accuracy=51.626302	 loss=2.049351	 lr=0.010000
Epoch[019] Batch [1249]/[3760]	Speed: 71.915495 samples/sec	 accuracy=51.557500	 loss=2.052996	 lr=0.010000
Epoch[019] Batch [1299]/[3760]	Speed: 71.172349 samples/sec	 accuracy=51.487981	 loss=2.055725	 lr=0.010000
Epoch[019] Batch [1349]/[3760]	Speed: 70.142649 samples/sec	 accuracy=51.440972	 loss=2.057296	 lr=0.010000
Epoch[019] Batch [1399]/[3760]	Speed: 68.197432 samples/sec	 accuracy=51.448661	 loss=2.056243	 lr=0.010000
Epoch[019] Batch [1449]/[3760]	Speed: 70.800033 samples/sec	 accuracy=51.431034	 loss=2.056380	 lr=0.010000
Epoch[019] Batch [1499]/[3760]	Speed: 69.543968 samples/sec	 accuracy=51.407292	 loss=2.058378	 lr=0.010000
Epoch[019] Batch [1549]/[3760]	Speed: 68.773444 samples/sec	 accuracy=51.417339	 loss=2.060029	 lr=0.010000
Epoch[019] Batch [1599]/[3760]	Speed: 70.991334 samples/sec	 accuracy=51.402344	 loss=2.060714	 lr=0.010000
Epoch[019] Batch [1649]/[3760]	Speed: 69.877715 samples/sec	 accuracy=51.406250	 loss=2.061163	 lr=0.010000
Epoch[019] Batch [1699]/[3760]	Speed: 68.596359 samples/sec	 accuracy=51.374081	 loss=2.063369	 lr=0.010000
Epoch[019] Batch [1749]/[3760]	Speed: 71.146124 samples/sec	 accuracy=51.337500	 loss=2.064799	 lr=0.010000
Epoch[019] Batch [1799]/[3760]	Speed: 69.810062 samples/sec	 accuracy=51.338542	 loss=2.065395	 lr=0.010000
Epoch[019] Batch [1849]/[3760]	Speed: 69.320780 samples/sec	 accuracy=51.350507	 loss=2.065096	 lr=0.010000
Epoch[019] Batch [1899]/[3760]	Speed: 68.791670 samples/sec	 accuracy=51.328125	 loss=2.065862	 lr=0.010000
Epoch[019] Batch [1949]/[3760]	Speed: 70.381175 samples/sec	 accuracy=51.331731	 loss=2.066603	 lr=0.010000
Epoch[019] Batch [1999]/[3760]	Speed: 70.179527 samples/sec	 accuracy=51.278125	 loss=2.068536	 lr=0.010000
Epoch[019] Batch [2049]/[3760]	Speed: 70.014984 samples/sec	 accuracy=51.286585	 loss=2.069257	 lr=0.010000
Epoch[019] Batch [2099]/[3760]	Speed: 69.763886 samples/sec	 accuracy=51.331101	 loss=2.067891	 lr=0.010000
Epoch[019] Batch [2149]/[3760]	Speed: 69.053746 samples/sec	 accuracy=51.309593	 loss=2.069426	 lr=0.010000
Epoch[019] Batch [2199]/[3760]	Speed: 70.100861 samples/sec	 accuracy=51.308239	 loss=2.068794	 lr=0.010000
Epoch[019] Batch [2249]/[3760]	Speed: 70.376194 samples/sec	 accuracy=51.286806	 loss=2.069273	 lr=0.010000
Epoch[019] Batch [2299]/[3760]	Speed: 69.340236 samples/sec	 accuracy=51.325408	 loss=2.068411	 lr=0.010000
Epoch[019] Batch [2349]/[3760]	Speed: 69.151065 samples/sec	 accuracy=51.309840	 loss=2.068641	 lr=0.010000
Epoch[019] Batch [2399]/[3760]	Speed: 70.842887 samples/sec	 accuracy=51.309245	 loss=2.069712	 lr=0.010000
Epoch[019] Batch [2449]/[3760]	Speed: 69.929252 samples/sec	 accuracy=51.310587	 loss=2.069734	 lr=0.010000
Epoch[019] Batch [2499]/[3760]	Speed: 69.097771 samples/sec	 accuracy=51.306875	 loss=2.069915	 lr=0.010000
Epoch[019] Batch [2549]/[3760]	Speed: 68.615490 samples/sec	 accuracy=51.295343	 loss=2.069219	 lr=0.010000
Epoch[019] Batch [2599]/[3760]	Speed: 70.595342 samples/sec	 accuracy=51.310697	 loss=2.068436	 lr=0.010000
Epoch[019] Batch [2649]/[3760]	Speed: 69.931247 samples/sec	 accuracy=51.303066	 loss=2.069095	 lr=0.010000
Epoch[019] Batch [2699]/[3760]	Speed: 70.458441 samples/sec	 accuracy=51.300347	 loss=2.068859	 lr=0.010000
Epoch[019] Batch [2749]/[3760]	Speed: 69.320827 samples/sec	 accuracy=51.297727	 loss=2.069754	 lr=0.010000
Epoch[019] Batch [2799]/[3760]	Speed: 69.093720 samples/sec	 accuracy=51.294643	 loss=2.070208	 lr=0.010000
Epoch[019] Batch [2849]/[3760]	Speed: 69.597157 samples/sec	 accuracy=51.320175	 loss=2.069996	 lr=0.010000
Epoch[019] Batch [2899]/[3760]	Speed: 70.049504 samples/sec	 accuracy=51.340517	 loss=2.068999	 lr=0.010000
Epoch[019] Batch [2949]/[3760]	Speed: 70.152924 samples/sec	 accuracy=51.349047	 loss=2.069519	 lr=0.010000
Epoch[019] Batch [2999]/[3760]	Speed: 69.671529 samples/sec	 accuracy=51.341146	 loss=2.069724	 lr=0.010000
Epoch[019] Batch [3049]/[3760]	Speed: 68.898139 samples/sec	 accuracy=51.325820	 loss=2.070271	 lr=0.010000
Epoch[019] Batch [3099]/[3760]	Speed: 69.980402 samples/sec	 accuracy=51.329637	 loss=2.070705	 lr=0.010000
Epoch[019] Batch [3149]/[3760]	Speed: 70.143676 samples/sec	 accuracy=51.324901	 loss=2.071292	 lr=0.010000
Epoch[019] Batch [3199]/[3760]	Speed: 69.994286 samples/sec	 accuracy=51.356445	 loss=2.070521	 lr=0.010000
Epoch[019] Batch [3249]/[3760]	Speed: 70.134155 samples/sec	 accuracy=51.352404	 loss=2.070560	 lr=0.010000
Epoch[019] Batch [3299]/[3760]	Speed: 68.180403 samples/sec	 accuracy=51.359375	 loss=2.070712	 lr=0.010000
Epoch[019] Batch [3349]/[3760]	Speed: 69.913800 samples/sec	 accuracy=51.353078	 loss=2.071164	 lr=0.010000
Epoch[019] Batch [3399]/[3760]	Speed: 69.400606 samples/sec	 accuracy=51.312500	 loss=2.072518	 lr=0.010000
Epoch[019] Batch [3449]/[3760]	Speed: 70.554457 samples/sec	 accuracy=51.308877	 loss=2.072930	 lr=0.010000
Epoch[019] Batch [3499]/[3760]	Speed: 69.061217 samples/sec	 accuracy=51.300893	 loss=2.072795	 lr=0.010000
Epoch[019] Batch [3549]/[3760]	Speed: 70.179933 samples/sec	 accuracy=51.290933	 loss=2.073175	 lr=0.010000
Epoch[019] Batch [3599]/[3760]	Speed: 70.261658 samples/sec	 accuracy=51.278212	 loss=2.074177	 lr=0.010000
Epoch[019] Batch [3649]/[3760]	Speed: 68.168859 samples/sec	 accuracy=51.263699	 loss=2.074648	 lr=0.010000
Epoch[019] Batch [3699]/[3760]	Speed: 69.694689 samples/sec	 accuracy=51.239865	 loss=2.075485	 lr=0.010000
Epoch[019] Batch [3749]/[3760]	Speed: 78.783679 samples/sec	 accuracy=51.232500	 loss=2.076031	 lr=0.010000
Batch [0049]/[0303]: acc-top1=52.156250 acc-top5=76.500000
Batch [0099]/[0303]: acc-top1=51.906250 acc-top5=76.843750
Batch [0149]/[0303]: acc-top1=52.687500 acc-top5=77.333333
Batch [0199]/[0303]: acc-top1=53.078125 acc-top5=77.484375
Batch [0249]/[0303]: acc-top1=52.962500 acc-top5=77.668750
Batch [0299]/[0303]: acc-top1=52.677083 acc-top5=77.411458
[Epoch 019] training: accuracy=51.232547	 loss=2.076031
[Epoch 019] speed: 69 samples/sec	time cost: 3757.755769
[Epoch 019] validation: acc-top1=52.645421 acc-top5=77.403053 loss=2.126544
Epoch[020] Batch [0049]/[3759]	Speed: 48.197884 samples/sec	 accuracy=52.531250	 loss=2.015838	 lr=0.010000
Epoch[020] Batch [0099]/[3759]	Speed: 66.763986 samples/sec	 accuracy=52.109375	 loss=2.039605	 lr=0.010000
Epoch[020] Batch [0149]/[3759]	Speed: 69.615434 samples/sec	 accuracy=52.260417	 loss=2.034919	 lr=0.010000
Epoch[020] Batch [0199]/[3759]	Speed: 70.096894 samples/sec	 accuracy=52.375000	 loss=2.032937	 lr=0.010000
Epoch[020] Batch [0249]/[3759]	Speed: 67.854228 samples/sec	 accuracy=52.287500	 loss=2.028490	 lr=0.010000
Epoch[020] Batch [0299]/[3759]	Speed: 71.418039 samples/sec	 accuracy=52.072917	 loss=2.031109	 lr=0.010000
Epoch[020] Batch [0349]/[3759]	Speed: 70.547835 samples/sec	 accuracy=52.169643	 loss=2.032104	 lr=0.010000
Epoch[020] Batch [0399]/[3759]	Speed: 69.072576 samples/sec	 accuracy=51.902344	 loss=2.043552	 lr=0.010000
Epoch[020] Batch [0449]/[3759]	Speed: 70.548879 samples/sec	 accuracy=51.819444	 loss=2.046797	 lr=0.010000
Epoch[020] Batch [0499]/[3759]	Speed: 67.962082 samples/sec	 accuracy=51.878125	 loss=2.044137	 lr=0.010000
Epoch[020] Batch [0549]/[3759]	Speed: 70.368560 samples/sec	 accuracy=51.940341	 loss=2.041352	 lr=0.010000
Epoch[020] Batch [0599]/[3759]	Speed: 70.159807 samples/sec	 accuracy=51.916667	 loss=2.043011	 lr=0.010000
Epoch[020] Batch [0649]/[3759]	Speed: 67.283519 samples/sec	 accuracy=51.911058	 loss=2.042793	 lr=0.010000
Epoch[020] Batch [0699]/[3759]	Speed: 70.537500 samples/sec	 accuracy=51.937500	 loss=2.042910	 lr=0.010000
Epoch[020] Batch [0749]/[3759]	Speed: 69.747595 samples/sec	 accuracy=51.872917	 loss=2.044741	 lr=0.010000
Epoch[020] Batch [0799]/[3759]	Speed: 69.027200 samples/sec	 accuracy=51.746094	 loss=2.048301	 lr=0.010000
Epoch[020] Batch [0849]/[3759]	Speed: 69.172607 samples/sec	 accuracy=51.707721	 loss=2.049293	 lr=0.010000
Epoch[020] Batch [0899]/[3759]	Speed: 69.972545 samples/sec	 accuracy=51.670139	 loss=2.050871	 lr=0.010000
Epoch[020] Batch [0949]/[3759]	Speed: 70.463250 samples/sec	 accuracy=51.679276	 loss=2.050905	 lr=0.010000
Epoch[020] Batch [0999]/[3759]	Speed: 69.122532 samples/sec	 accuracy=51.651563	 loss=2.051840	 lr=0.010000
Epoch[020] Batch [1049]/[3759]	Speed: 68.518922 samples/sec	 accuracy=51.702381	 loss=2.049784	 lr=0.010000
Epoch[020] Batch [1099]/[3759]	Speed: 68.896987 samples/sec	 accuracy=51.725852	 loss=2.048222	 lr=0.010000
Epoch[020] Batch [1149]/[3759]	Speed: 71.658993 samples/sec	 accuracy=51.692935	 loss=2.048952	 lr=0.010000
Epoch[020] Batch [1199]/[3759]	Speed: 69.431596 samples/sec	 accuracy=51.697917	 loss=2.049268	 lr=0.010000
Epoch[020] Batch [1249]/[3759]	Speed: 69.179939 samples/sec	 accuracy=51.667500	 loss=2.049698	 lr=0.010000
Epoch[020] Batch [1299]/[3759]	Speed: 69.663277 samples/sec	 accuracy=51.667067	 loss=2.050585	 lr=0.010000
Epoch[020] Batch [1349]/[3759]	Speed: 70.083089 samples/sec	 accuracy=51.668981	 loss=2.050558	 lr=0.010000
Epoch[020] Batch [1399]/[3759]	Speed: 69.235402 samples/sec	 accuracy=51.731027	 loss=2.048558	 lr=0.010000
Epoch[020] Batch [1449]/[3759]	Speed: 70.626314 samples/sec	 accuracy=51.752155	 loss=2.048962	 lr=0.010000
Epoch[020] Batch [1499]/[3759]	Speed: 70.672660 samples/sec	 accuracy=51.784375	 loss=2.047474	 lr=0.010000
Epoch[020] Batch [1549]/[3759]	Speed: 68.134743 samples/sec	 accuracy=51.783266	 loss=2.048293	 lr=0.010000
Epoch[020] Batch [1599]/[3759]	Speed: 69.760266 samples/sec	 accuracy=51.751953	 loss=2.050058	 lr=0.010000
Epoch[020] Batch [1649]/[3759]	Speed: 70.255900 samples/sec	 accuracy=51.753788	 loss=2.049959	 lr=0.010000
Epoch[020] Batch [1699]/[3759]	Speed: 68.852427 samples/sec	 accuracy=51.741728	 loss=2.051037	 lr=0.010000
Epoch[020] Batch [1749]/[3759]	Speed: 69.661909 samples/sec	 accuracy=51.750893	 loss=2.051182	 lr=0.010000
Epoch[020] Batch [1799]/[3759]	Speed: 71.017067 samples/sec	 accuracy=51.724826	 loss=2.051949	 lr=0.010000
Epoch[020] Batch [1849]/[3759]	Speed: 70.124619 samples/sec	 accuracy=51.680743	 loss=2.053012	 lr=0.010000
Epoch[020] Batch [1899]/[3759]	Speed: 69.058357 samples/sec	 accuracy=51.658717	 loss=2.053793	 lr=0.010000
Epoch[020] Batch [1949]/[3759]	Speed: 68.803538 samples/sec	 accuracy=51.621795	 loss=2.055674	 lr=0.010000
Epoch[020] Batch [1999]/[3759]	Speed: 69.860428 samples/sec	 accuracy=51.614844	 loss=2.055604	 lr=0.010000
Epoch[020] Batch [2049]/[3759]	Speed: 69.717747 samples/sec	 accuracy=51.613567	 loss=2.055261	 lr=0.010000
Epoch[020] Batch [2099]/[3759]	Speed: 69.104226 samples/sec	 accuracy=51.590030	 loss=2.055666	 lr=0.010000
Epoch[020] Batch [2149]/[3759]	Speed: 70.705879 samples/sec	 accuracy=51.625727	 loss=2.056050	 lr=0.010000
Epoch[020] Batch [2199]/[3759]	Speed: 70.147208 samples/sec	 accuracy=51.634943	 loss=2.056481	 lr=0.010000
Epoch[020] Batch [2249]/[3759]	Speed: 68.910368 samples/sec	 accuracy=51.661111	 loss=2.055512	 lr=0.010000
Epoch[020] Batch [2299]/[3759]	Speed: 68.914502 samples/sec	 accuracy=51.639266	 loss=2.056839	 lr=0.010000
Epoch[020] Batch [2349]/[3759]	Speed: 70.547825 samples/sec	 accuracy=51.616356	 loss=2.057885	 lr=0.010000
Epoch[020] Batch [2399]/[3759]	Speed: 69.070842 samples/sec	 accuracy=51.641276	 loss=2.056523	 lr=0.010000
Epoch[020] Batch [2449]/[3759]	Speed: 69.223043 samples/sec	 accuracy=51.663265	 loss=2.055710	 lr=0.010000
Epoch[020] Batch [2499]/[3759]	Speed: 70.606222 samples/sec	 accuracy=51.644375	 loss=2.056713	 lr=0.010000
Epoch[020] Batch [2549]/[3759]	Speed: 70.393540 samples/sec	 accuracy=51.634191	 loss=2.057093	 lr=0.010000
Epoch[020] Batch [2599]/[3759]	Speed: 67.569887 samples/sec	 accuracy=51.641226	 loss=2.057455	 lr=0.010000
Epoch[020] Batch [2649]/[3759]	Speed: 70.679855 samples/sec	 accuracy=51.625000	 loss=2.057848	 lr=0.010000
Epoch[020] Batch [2699]/[3759]	Speed: 70.459789 samples/sec	 accuracy=51.608796	 loss=2.058543	 lr=0.010000
Epoch[020] Batch [2749]/[3759]	Speed: 69.851135 samples/sec	 accuracy=51.606818	 loss=2.057788	 lr=0.010000
Epoch[020] Batch [2799]/[3759]	Speed: 68.386835 samples/sec	 accuracy=51.599330	 loss=2.058005	 lr=0.010000
Epoch[020] Batch [2849]/[3759]	Speed: 70.442808 samples/sec	 accuracy=51.607456	 loss=2.058337	 lr=0.010000
Epoch[020] Batch [2899]/[3759]	Speed: 68.667726 samples/sec	 accuracy=51.626078	 loss=2.058285	 lr=0.010000
Epoch[020] Batch [2949]/[3759]	Speed: 70.291141 samples/sec	 accuracy=51.602225	 loss=2.058732	 lr=0.010000
Epoch[020] Batch [2999]/[3759]	Speed: 70.087731 samples/sec	 accuracy=51.588021	 loss=2.059302	 lr=0.010000
Epoch[020] Batch [3049]/[3759]	Speed: 68.666141 samples/sec	 accuracy=51.581967	 loss=2.059248	 lr=0.010000
Epoch[020] Batch [3099]/[3759]	Speed: 71.274224 samples/sec	 accuracy=51.574093	 loss=2.059394	 lr=0.010000
Epoch[020] Batch [3149]/[3759]	Speed: 69.234196 samples/sec	 accuracy=51.557540	 loss=2.059724	 lr=0.010000
Epoch[020] Batch [3199]/[3759]	Speed: 70.578629 samples/sec	 accuracy=51.545898	 loss=2.060346	 lr=0.010000
Epoch[020] Batch [3249]/[3759]	Speed: 69.441846 samples/sec	 accuracy=51.541827	 loss=2.060563	 lr=0.010000
Epoch[020] Batch [3299]/[3759]	Speed: 69.088196 samples/sec	 accuracy=51.551136	 loss=2.061266	 lr=0.010000
Epoch[020] Batch [3349]/[3759]	Speed: 70.182516 samples/sec	 accuracy=51.563899	 loss=2.060491	 lr=0.010000
Epoch[020] Batch [3399]/[3759]	Speed: 68.791146 samples/sec	 accuracy=51.558364	 loss=2.060721	 lr=0.010000
Epoch[020] Batch [3449]/[3759]	Speed: 69.935256 samples/sec	 accuracy=51.574728	 loss=2.059568	 lr=0.010000
Epoch[020] Batch [3499]/[3759]	Speed: 68.782243 samples/sec	 accuracy=51.570536	 loss=2.059475	 lr=0.010000
Epoch[020] Batch [3549]/[3759]	Speed: 70.730106 samples/sec	 accuracy=51.573504	 loss=2.059493	 lr=0.010000
Epoch[020] Batch [3599]/[3759]	Speed: 69.465633 samples/sec	 accuracy=51.576389	 loss=2.059538	 lr=0.010000
Epoch[020] Batch [3649]/[3759]	Speed: 69.936834 samples/sec	 accuracy=51.585616	 loss=2.059752	 lr=0.010000
Epoch[020] Batch [3699]/[3759]	Speed: 69.776958 samples/sec	 accuracy=51.568412	 loss=2.060267	 lr=0.010000
Epoch[020] Batch [3749]/[3759]	Speed: 77.872907 samples/sec	 accuracy=51.550000	 loss=2.061443	 lr=0.010000
Batch [0049]/[0303]: acc-top1=52.718750 acc-top5=77.718750
Batch [0099]/[0303]: acc-top1=52.531250 acc-top5=77.781250
Batch [0149]/[0303]: acc-top1=53.020833 acc-top5=78.250000
Batch [0199]/[0303]: acc-top1=53.101562 acc-top5=78.335938
Batch [0249]/[0303]: acc-top1=53.225000 acc-top5=78.456250
Batch [0299]/[0303]: acc-top1=53.026042 acc-top5=78.177083
[Epoch 020] training: accuracy=51.559590	 loss=2.061152
[Epoch 020] speed: 69 samples/sec	time cost: 3763.495981
[Epoch 020] validation: acc-top1=53.021865 acc-top5=78.150784 loss=2.079734
Epoch[021] Batch [0049]/[3760]	Speed: 47.902668 samples/sec	 accuracy=53.375000	 loss=1.992168	 lr=0.010000
Epoch[021] Batch [0099]/[3760]	Speed: 70.087939 samples/sec	 accuracy=53.312500	 loss=1.994283	 lr=0.010000
Epoch[021] Batch [0149]/[3760]	Speed: 67.663662 samples/sec	 accuracy=52.739583	 loss=2.012951	 lr=0.010000
Epoch[021] Batch [0199]/[3760]	Speed: 71.045141 samples/sec	 accuracy=52.289063	 loss=2.024684	 lr=0.010000
Epoch[021] Batch [0249]/[3760]	Speed: 69.779605 samples/sec	 accuracy=52.387500	 loss=2.028570	 lr=0.010000
Epoch[021] Batch [0299]/[3760]	Speed: 69.512825 samples/sec	 accuracy=52.260417	 loss=2.033339	 lr=0.010000
Epoch[021] Batch [0349]/[3760]	Speed: 68.858633 samples/sec	 accuracy=52.263393	 loss=2.035187	 lr=0.010000
Epoch[021] Batch [0399]/[3760]	Speed: 70.261444 samples/sec	 accuracy=52.292969	 loss=2.030754	 lr=0.010000
Epoch[021] Batch [0449]/[3760]	Speed: 67.839017 samples/sec	 accuracy=52.267361	 loss=2.032483	 lr=0.010000
Epoch[021] Batch [0499]/[3760]	Speed: 71.876387 samples/sec	 accuracy=52.296875	 loss=2.030401	 lr=0.010000
Epoch[021] Batch [0549]/[3760]	Speed: 69.440068 samples/sec	 accuracy=52.346591	 loss=2.028626	 lr=0.010000
Epoch[021] Batch [0599]/[3760]	Speed: 69.902284 samples/sec	 accuracy=52.335937	 loss=2.027099	 lr=0.010000
Epoch[021] Batch [0649]/[3760]	Speed: 68.544066 samples/sec	 accuracy=52.240385	 loss=2.028531	 lr=0.010000
Epoch[021] Batch [0699]/[3760]	Speed: 69.830149 samples/sec	 accuracy=52.183036	 loss=2.028873	 lr=0.010000
Epoch[021] Batch [0749]/[3760]	Speed: 68.599095 samples/sec	 accuracy=52.164583	 loss=2.028331	 lr=0.010000
Epoch[021] Batch [0799]/[3760]	Speed: 68.752891 samples/sec	 accuracy=52.154297	 loss=2.026037	 lr=0.010000
Epoch[021] Batch [0849]/[3760]	Speed: 71.056911 samples/sec	 accuracy=52.170956	 loss=2.024379	 lr=0.010000
Epoch[021] Batch [0899]/[3760]	Speed: 69.135802 samples/sec	 accuracy=51.920139	 loss=2.034618	 lr=0.010000
Epoch[021] Batch [0949]/[3760]	Speed: 69.213366 samples/sec	 accuracy=51.921053	 loss=2.034244	 lr=0.010000
Epoch[021] Batch [0999]/[3760]	Speed: 70.623108 samples/sec	 accuracy=51.893750	 loss=2.034805	 lr=0.010000
Epoch[021] Batch [1049]/[3760]	Speed: 69.999774 samples/sec	 accuracy=51.873512	 loss=2.035120	 lr=0.010000
Epoch[021] Batch [1099]/[3760]	Speed: 68.599508 samples/sec	 accuracy=51.867898	 loss=2.036541	 lr=0.010000
Epoch[021] Batch [1149]/[3760]	Speed: 69.487612 samples/sec	 accuracy=51.830163	 loss=2.039112	 lr=0.010000
Epoch[021] Batch [1199]/[3760]	Speed: 70.330036 samples/sec	 accuracy=51.774740	 loss=2.039915	 lr=0.010000
Epoch[021] Batch [1249]/[3760]	Speed: 70.331399 samples/sec	 accuracy=51.796250	 loss=2.039753	 lr=0.010000
Epoch[021] Batch [1299]/[3760]	Speed: 68.263169 samples/sec	 accuracy=51.750000	 loss=2.040254	 lr=0.010000
Epoch[021] Batch [1349]/[3760]	Speed: 71.117440 samples/sec	 accuracy=51.759259	 loss=2.040675	 lr=0.010000
Epoch[021] Batch [1399]/[3760]	Speed: 66.236495 samples/sec	 accuracy=51.755580	 loss=2.039030	 lr=0.010000
Epoch[021] Batch [1449]/[3760]	Speed: 72.178255 samples/sec	 accuracy=51.701509	 loss=2.040919	 lr=0.010000
Epoch[021] Batch [1499]/[3760]	Speed: 69.032591 samples/sec	 accuracy=51.750000	 loss=2.040067	 lr=0.010000
Epoch[021] Batch [1549]/[3760]	Speed: 71.384389 samples/sec	 accuracy=51.771169	 loss=2.039556	 lr=0.010000
Epoch[021] Batch [1599]/[3760]	Speed: 69.933670 samples/sec	 accuracy=51.827148	 loss=2.036382	 lr=0.010000
Epoch[021] Batch [1649]/[3760]	Speed: 67.396563 samples/sec	 accuracy=51.817235	 loss=2.035645	 lr=0.010000
Epoch[021] Batch [1699]/[3760]	Speed: 71.289513 samples/sec	 accuracy=51.804228	 loss=2.035461	 lr=0.010000
Epoch[021] Batch [1749]/[3760]	Speed: 67.776877 samples/sec	 accuracy=51.808036	 loss=2.036087	 lr=0.010000
Epoch[021] Batch [1799]/[3760]	Speed: 70.857262 samples/sec	 accuracy=51.804688	 loss=2.036286	 lr=0.010000
Epoch[021] Batch [1849]/[3760]	Speed: 66.564233 samples/sec	 accuracy=51.856419	 loss=2.034682	 lr=0.010000
Epoch[021] Batch [1899]/[3760]	Speed: 71.609476 samples/sec	 accuracy=51.838816	 loss=2.036074	 lr=0.010000
Epoch[021] Batch [1949]/[3760]	Speed: 70.454214 samples/sec	 accuracy=51.838942	 loss=2.036169	 lr=0.010000
Epoch[021] Batch [1999]/[3760]	Speed: 68.954531 samples/sec	 accuracy=51.837500	 loss=2.037126	 lr=0.010000
Epoch[021] Batch [2049]/[3760]	Speed: 69.133329 samples/sec	 accuracy=51.827744	 loss=2.038098	 lr=0.010000
Epoch[021] Batch [2099]/[3760]	Speed: 69.994063 samples/sec	 accuracy=51.817708	 loss=2.038781	 lr=0.010000
Epoch[021] Batch [2149]/[3760]	Speed: 69.138541 samples/sec	 accuracy=51.805233	 loss=2.039087	 lr=0.010000
Epoch[021] Batch [2199]/[3760]	Speed: 70.585976 samples/sec	 accuracy=51.773438	 loss=2.040242	 lr=0.010000
Epoch[021] Batch [2249]/[3760]	Speed: 69.107554 samples/sec	 accuracy=51.786806	 loss=2.039872	 lr=0.010000
Epoch[021] Batch [2299]/[3760]	Speed: 70.103776 samples/sec	 accuracy=51.767663	 loss=2.040555	 lr=0.010000
Epoch[021] Batch [2349]/[3760]	Speed: 69.793819 samples/sec	 accuracy=51.771941	 loss=2.041090	 lr=0.010000
Epoch[021] Batch [2399]/[3760]	Speed: 69.506602 samples/sec	 accuracy=51.755859	 loss=2.041135	 lr=0.010000
Epoch[021] Batch [2449]/[3760]	Speed: 69.079497 samples/sec	 accuracy=51.770408	 loss=2.041389	 lr=0.010000
Epoch[021] Batch [2499]/[3760]	Speed: 70.200997 samples/sec	 accuracy=51.746250	 loss=2.043251	 lr=0.010000
Epoch[021] Batch [2549]/[3760]	Speed: 69.358328 samples/sec	 accuracy=51.752451	 loss=2.042790	 lr=0.010000
Epoch[021] Batch [2599]/[3760]	Speed: 69.281215 samples/sec	 accuracy=51.745793	 loss=2.042696	 lr=0.010000
Epoch[021] Batch [2649]/[3760]	Speed: 70.505096 samples/sec	 accuracy=51.741745	 loss=2.042423	 lr=0.010000
Epoch[021] Batch [2699]/[3760]	Speed: 68.905586 samples/sec	 accuracy=51.760995	 loss=2.041533	 lr=0.010000
Epoch[021] Batch [2749]/[3760]	Speed: 70.248563 samples/sec	 accuracy=51.748295	 loss=2.042340	 lr=0.010000
Epoch[021] Batch [2799]/[3760]	Speed: 71.217481 samples/sec	 accuracy=51.728795	 loss=2.043172	 lr=0.010000
Epoch[021] Batch [2849]/[3760]	Speed: 67.215630 samples/sec	 accuracy=51.741228	 loss=2.042594	 lr=0.010000
Epoch[021] Batch [2899]/[3760]	Speed: 71.486296 samples/sec	 accuracy=51.764009	 loss=2.042386	 lr=0.010000
Epoch[021] Batch [2949]/[3760]	Speed: 68.427184 samples/sec	 accuracy=51.734640	 loss=2.042983	 lr=0.010000
Epoch[021] Batch [2999]/[3760]	Speed: 70.300698 samples/sec	 accuracy=51.752604	 loss=2.042634	 lr=0.010000
Epoch[021] Batch [3049]/[3760]	Speed: 70.010921 samples/sec	 accuracy=51.759734	 loss=2.043153	 lr=0.010000
Epoch[021] Batch [3099]/[3760]	Speed: 68.609861 samples/sec	 accuracy=51.759577	 loss=2.044203	 lr=0.010000
Epoch[021] Batch [3149]/[3760]	Speed: 69.165560 samples/sec	 accuracy=51.738591	 loss=2.045976	 lr=0.010000
Epoch[021] Batch [3199]/[3760]	Speed: 70.597717 samples/sec	 accuracy=51.713867	 loss=2.047195	 lr=0.010000
Epoch[021] Batch [3249]/[3760]	Speed: 69.651801 samples/sec	 accuracy=51.688942	 loss=2.047871	 lr=0.010000
Epoch[021] Batch [3299]/[3760]	Speed: 69.001878 samples/sec	 accuracy=51.681818	 loss=2.048525	 lr=0.010000
Epoch[021] Batch [3349]/[3760]	Speed: 68.803590 samples/sec	 accuracy=51.666978	 loss=2.048435	 lr=0.010000
Epoch[021] Batch [3399]/[3760]	Speed: 68.560571 samples/sec	 accuracy=51.656250	 loss=2.048908	 lr=0.010000
Epoch[021] Batch [3449]/[3760]	Speed: 70.589378 samples/sec	 accuracy=51.669384	 loss=2.048604	 lr=0.010000
Epoch[021] Batch [3499]/[3760]	Speed: 71.046564 samples/sec	 accuracy=51.666964	 loss=2.049111	 lr=0.010000
Epoch[021] Batch [3549]/[3760]	Speed: 68.438679 samples/sec	 accuracy=51.663292	 loss=2.049893	 lr=0.010000
Epoch[021] Batch [3599]/[3760]	Speed: 70.027371 samples/sec	 accuracy=51.650608	 loss=2.050903	 lr=0.010000
Epoch[021] Batch [3649]/[3760]	Speed: 71.606379 samples/sec	 accuracy=51.643408	 loss=2.051171	 lr=0.010000
Epoch[021] Batch [3699]/[3760]	Speed: 69.661781 samples/sec	 accuracy=51.632601	 loss=2.051441	 lr=0.010000
Epoch[021] Batch [3749]/[3760]	Speed: 76.391690 samples/sec	 accuracy=51.639167	 loss=2.051888	 lr=0.010000
Batch [0049]/[0303]: acc-top1=51.812500 acc-top5=76.531250
Batch [0099]/[0303]: acc-top1=51.390625 acc-top5=76.406250
Batch [0149]/[0303]: acc-top1=52.291667 acc-top5=77.166667
Batch [0199]/[0303]: acc-top1=52.453125 acc-top5=77.328125
Batch [0249]/[0303]: acc-top1=52.350000 acc-top5=77.381250
Batch [0299]/[0303]: acc-top1=52.067708 acc-top5=77.125000
[Epoch 021] training: accuracy=51.629405	 loss=2.052442
[Epoch 021] speed: 69 samples/sec	time cost: 3765.057517
[Epoch 021] validation: acc-top1=52.057550 acc-top5=77.083333 loss=2.137917
Epoch[022] Batch [0049]/[3760]	Speed: 47.928560 samples/sec	 accuracy=52.781250	 loss=2.002576	 lr=0.010000
Epoch[022] Batch [0099]/[3760]	Speed: 69.879135 samples/sec	 accuracy=52.578125	 loss=2.002104	 lr=0.010000
Epoch[022] Batch [0149]/[3760]	Speed: 68.700481 samples/sec	 accuracy=52.760417	 loss=2.005243	 lr=0.010000
Epoch[022] Batch [0199]/[3760]	Speed: 67.678101 samples/sec	 accuracy=53.109375	 loss=1.997350	 lr=0.010000
Epoch[022] Batch [0249]/[3760]	Speed: 69.119636 samples/sec	 accuracy=53.112500	 loss=1.990179	 lr=0.010000
Epoch[022] Batch [0299]/[3760]	Speed: 70.959874 samples/sec	 accuracy=53.005208	 loss=1.998233	 lr=0.010000
Epoch[022] Batch [0349]/[3760]	Speed: 70.159499 samples/sec	 accuracy=53.183036	 loss=1.995316	 lr=0.010000
Epoch[022] Batch [0399]/[3760]	Speed: 70.249090 samples/sec	 accuracy=53.179687	 loss=1.995173	 lr=0.010000
Epoch[022] Batch [0449]/[3760]	Speed: 69.206088 samples/sec	 accuracy=53.194444	 loss=1.996535	 lr=0.010000
Epoch[022] Batch [0499]/[3760]	Speed: 70.524538 samples/sec	 accuracy=53.118750	 loss=1.999223	 lr=0.010000
Epoch[022] Batch [0549]/[3760]	Speed: 69.087741 samples/sec	 accuracy=52.980114	 loss=2.004028	 lr=0.010000
Epoch[022] Batch [0599]/[3760]	Speed: 69.621809 samples/sec	 accuracy=52.851563	 loss=2.007177	 lr=0.010000
Epoch[022] Batch [0649]/[3760]	Speed: 70.486947 samples/sec	 accuracy=52.841346	 loss=2.004106	 lr=0.010000
Epoch[022] Batch [0699]/[3760]	Speed: 69.366471 samples/sec	 accuracy=52.828125	 loss=2.003824	 lr=0.010000
Epoch[022] Batch [0749]/[3760]	Speed: 68.672244 samples/sec	 accuracy=52.812500	 loss=2.002379	 lr=0.010000
Epoch[022] Batch [0799]/[3760]	Speed: 69.304939 samples/sec	 accuracy=52.806641	 loss=2.002577	 lr=0.010000
Epoch[022] Batch [0849]/[3760]	Speed: 71.207676 samples/sec	 accuracy=52.838235	 loss=2.000633	 lr=0.010000
Epoch[022] Batch [0899]/[3760]	Speed: 70.106671 samples/sec	 accuracy=52.690972	 loss=2.005391	 lr=0.010000
Epoch[022] Batch [0949]/[3760]	Speed: 68.648462 samples/sec	 accuracy=52.539474	 loss=2.011538	 lr=0.010000
Epoch[022] Batch [0999]/[3760]	Speed: 69.460752 samples/sec	 accuracy=52.490625	 loss=2.013348	 lr=0.010000
Epoch[022] Batch [1049]/[3760]	Speed: 70.594009 samples/sec	 accuracy=52.535714	 loss=2.013228	 lr=0.010000
Epoch[022] Batch [1099]/[3760]	Speed: 70.596901 samples/sec	 accuracy=52.414773	 loss=2.017085	 lr=0.010000
Epoch[022] Batch [1149]/[3760]	Speed: 69.982473 samples/sec	 accuracy=52.361413	 loss=2.016745	 lr=0.010000
Epoch[022] Batch [1199]/[3760]	Speed: 69.307785 samples/sec	 accuracy=52.348958	 loss=2.016931	 lr=0.010000
Epoch[022] Batch [1249]/[3760]	Speed: 70.126620 samples/sec	 accuracy=52.356250	 loss=2.018666	 lr=0.010000
Epoch[022] Batch [1299]/[3760]	Speed: 71.735734 samples/sec	 accuracy=52.308894	 loss=2.021600	 lr=0.010000
Epoch[022] Batch [1349]/[3760]	Speed: 68.729038 samples/sec	 accuracy=52.281250	 loss=2.021958	 lr=0.010000
Epoch[022] Batch [1399]/[3760]	Speed: 67.382752 samples/sec	 accuracy=52.286830	 loss=2.022405	 lr=0.010000
Epoch[022] Batch [1449]/[3760]	Speed: 70.272612 samples/sec	 accuracy=52.284483	 loss=2.022497	 lr=0.010000
Epoch[022] Batch [1499]/[3760]	Speed: 69.746844 samples/sec	 accuracy=52.296875	 loss=2.022691	 lr=0.010000
Epoch[022] Batch [1549]/[3760]	Speed: 70.665069 samples/sec	 accuracy=52.291331	 loss=2.024465	 lr=0.010000
Epoch[022] Batch [1599]/[3760]	Speed: 69.739750 samples/sec	 accuracy=52.339844	 loss=2.023118	 lr=0.010000
Epoch[022] Batch [1649]/[3760]	Speed: 69.902987 samples/sec	 accuracy=52.292614	 loss=2.024640	 lr=0.010000
Epoch[022] Batch [1699]/[3760]	Speed: 67.702435 samples/sec	 accuracy=52.199449	 loss=2.027921	 lr=0.010000
Epoch[022] Batch [1749]/[3760]	Speed: 70.321394 samples/sec	 accuracy=52.164286	 loss=2.029593	 lr=0.010000
Epoch[022] Batch [1799]/[3760]	Speed: 69.566550 samples/sec	 accuracy=52.125868	 loss=2.031985	 lr=0.010000
Epoch[022] Batch [1849]/[3760]	Speed: 69.661597 samples/sec	 accuracy=52.116554	 loss=2.032634	 lr=0.010000
Epoch[022] Batch [1899]/[3760]	Speed: 70.799896 samples/sec	 accuracy=52.105263	 loss=2.032778	 lr=0.010000
Epoch[022] Batch [1949]/[3760]	Speed: 69.045718 samples/sec	 accuracy=52.092949	 loss=2.033643	 lr=0.010000
Epoch[022] Batch [1999]/[3760]	Speed: 68.763082 samples/sec	 accuracy=52.104688	 loss=2.032512	 lr=0.010000
Epoch[022] Batch [2049]/[3760]	Speed: 71.466618 samples/sec	 accuracy=52.080030	 loss=2.033089	 lr=0.010000
Epoch[022] Batch [2099]/[3760]	Speed: 69.492571 samples/sec	 accuracy=52.082589	 loss=2.033049	 lr=0.010000
Epoch[022] Batch [2149]/[3760]	Speed: 68.340304 samples/sec	 accuracy=52.120640	 loss=2.032967	 lr=0.010000
Epoch[022] Batch [2199]/[3760]	Speed: 70.038119 samples/sec	 accuracy=52.144886	 loss=2.032146	 lr=0.010000
Epoch[022] Batch [2249]/[3760]	Speed: 70.559820 samples/sec	 accuracy=52.154167	 loss=2.031734	 lr=0.010000
Epoch[022] Batch [2299]/[3760]	Speed: 68.140878 samples/sec	 accuracy=52.141304	 loss=2.031910	 lr=0.010000
Epoch[022] Batch [2349]/[3760]	Speed: 69.882151 samples/sec	 accuracy=52.152926	 loss=2.031732	 lr=0.010000
Epoch[022] Batch [2399]/[3760]	Speed: 70.213193 samples/sec	 accuracy=52.148438	 loss=2.031677	 lr=0.010000
Epoch[022] Batch [2449]/[3760]	Speed: 68.485121 samples/sec	 accuracy=52.135842	 loss=2.032980	 lr=0.010000
Epoch[022] Batch [2499]/[3760]	Speed: 68.555523 samples/sec	 accuracy=52.103750	 loss=2.033818	 lr=0.010000
Epoch[022] Batch [2549]/[3760]	Speed: 70.295286 samples/sec	 accuracy=52.126838	 loss=2.032953	 lr=0.010000
Epoch[022] Batch [2599]/[3760]	Speed: 70.397397 samples/sec	 accuracy=52.144832	 loss=2.032690	 lr=0.010000
Epoch[022] Batch [2649]/[3760]	Speed: 70.908127 samples/sec	 accuracy=52.137382	 loss=2.033255	 lr=0.010000
Epoch[022] Batch [2699]/[3760]	Speed: 69.769591 samples/sec	 accuracy=52.119792	 loss=2.034101	 lr=0.010000
Epoch[022] Batch [2749]/[3760]	Speed: 70.455244 samples/sec	 accuracy=52.149432	 loss=2.032998	 lr=0.010000
Epoch[022] Batch [2799]/[3760]	Speed: 69.908190 samples/sec	 accuracy=52.147879	 loss=2.032548	 lr=0.010000
Epoch[022] Batch [2849]/[3760]	Speed: 69.215694 samples/sec	 accuracy=52.154057	 loss=2.032690	 lr=0.010000
Epoch[022] Batch [2899]/[3760]	Speed: 68.396912 samples/sec	 accuracy=52.144397	 loss=2.032495	 lr=0.010000
Epoch[022] Batch [2949]/[3760]	Speed: 70.530892 samples/sec	 accuracy=52.134534	 loss=2.032832	 lr=0.010000
Epoch[022] Batch [2999]/[3760]	Speed: 70.446867 samples/sec	 accuracy=52.140104	 loss=2.032001	 lr=0.010000
Epoch[022] Batch [3049]/[3760]	Speed: 69.111747 samples/sec	 accuracy=52.144980	 loss=2.031679	 lr=0.010000
Epoch[022] Batch [3099]/[3760]	Speed: 69.421217 samples/sec	 accuracy=52.150202	 loss=2.031784	 lr=0.010000
Epoch[022] Batch [3149]/[3760]	Speed: 69.533817 samples/sec	 accuracy=52.141865	 loss=2.031876	 lr=0.010000
Epoch[022] Batch [3199]/[3760]	Speed: 70.301872 samples/sec	 accuracy=52.119629	 loss=2.032793	 lr=0.010000
Epoch[022] Batch [3249]/[3760]	Speed: 69.263386 samples/sec	 accuracy=52.101442	 loss=2.033915	 lr=0.010000
Epoch[022] Batch [3299]/[3760]	Speed: 68.571825 samples/sec	 accuracy=52.086174	 loss=2.034231	 lr=0.010000
Epoch[022] Batch [3349]/[3760]	Speed: 70.039493 samples/sec	 accuracy=52.068563	 loss=2.034247	 lr=0.010000
Epoch[022] Batch [3399]/[3760]	Speed: 70.717971 samples/sec	 accuracy=52.079504	 loss=2.033835	 lr=0.010000
Epoch[022] Batch [3449]/[3760]	Speed: 69.260664 samples/sec	 accuracy=52.089674	 loss=2.033650	 lr=0.010000
Epoch[022] Batch [3499]/[3760]	Speed: 69.362984 samples/sec	 accuracy=52.071875	 loss=2.034412	 lr=0.010000
Epoch[022] Batch [3549]/[3760]	Speed: 70.696123 samples/sec	 accuracy=52.066461	 loss=2.034937	 lr=0.010000
Epoch[022] Batch [3599]/[3760]	Speed: 67.369191 samples/sec	 accuracy=52.056424	 loss=2.035491	 lr=0.010000
Epoch[022] Batch [3649]/[3760]	Speed: 70.380478 samples/sec	 accuracy=52.077055	 loss=2.035175	 lr=0.010000
Epoch[022] Batch [3699]/[3760]	Speed: 69.767335 samples/sec	 accuracy=52.067145	 loss=2.035206	 lr=0.010000
Epoch[022] Batch [3749]/[3760]	Speed: 78.022354 samples/sec	 accuracy=52.072500	 loss=2.035222	 lr=0.010000
Batch [0049]/[0303]: acc-top1=52.656250 acc-top5=76.750000
Batch [0099]/[0303]: acc-top1=52.171875 acc-top5=77.296875
Batch [0149]/[0303]: acc-top1=52.833333 acc-top5=77.531250
Batch [0199]/[0303]: acc-top1=53.085938 acc-top5=77.757812
Batch [0249]/[0303]: acc-top1=53.018750 acc-top5=77.868750
Batch [0299]/[0303]: acc-top1=52.755208 acc-top5=77.697917
[Epoch 022] training: accuracy=52.069897	 loss=2.035448
[Epoch 022] speed: 69 samples/sec	time cost: 3760.937901
[Epoch 022] validation: acc-top1=52.702145 acc-top5=77.686675 loss=2.085401
Epoch[023] Batch [0049]/[3759]	Speed: 48.201487 samples/sec	 accuracy=52.750000	 loss=2.012517	 lr=0.010000
Epoch[023] Batch [0099]/[3759]	Speed: 70.010885 samples/sec	 accuracy=53.000000	 loss=2.003447	 lr=0.010000
Epoch[023] Batch [0149]/[3759]	Speed: 69.091703 samples/sec	 accuracy=52.656250	 loss=2.002013	 lr=0.010000
Epoch[023] Batch [0199]/[3759]	Speed: 68.903797 samples/sec	 accuracy=52.726562	 loss=1.989547	 lr=0.010000
Epoch[023] Batch [0249]/[3759]	Speed: 69.798065 samples/sec	 accuracy=52.868750	 loss=1.986780	 lr=0.010000
Epoch[023] Batch [0299]/[3759]	Speed: 70.843576 samples/sec	 accuracy=52.927083	 loss=1.994432	 lr=0.010000
Epoch[023] Batch [0349]/[3759]	Speed: 69.623601 samples/sec	 accuracy=52.714286	 loss=2.007622	 lr=0.010000
Epoch[023] Batch [0399]/[3759]	Speed: 69.752781 samples/sec	 accuracy=52.460938	 loss=2.012784	 lr=0.010000
Epoch[023] Batch [0449]/[3759]	Speed: 67.895249 samples/sec	 accuracy=52.406250	 loss=2.014715	 lr=0.010000
Epoch[023] Batch [0499]/[3759]	Speed: 70.185916 samples/sec	 accuracy=52.315625	 loss=2.018760	 lr=0.010000
Epoch[023] Batch [0549]/[3759]	Speed: 69.932289 samples/sec	 accuracy=52.272727	 loss=2.024038	 lr=0.010000
Epoch[023] Batch [0599]/[3759]	Speed: 70.335284 samples/sec	 accuracy=52.348958	 loss=2.020205	 lr=0.010000
Epoch[023] Batch [0649]/[3759]	Speed: 68.504390 samples/sec	 accuracy=52.401442	 loss=2.018503	 lr=0.010000
Epoch[023] Batch [0699]/[3759]	Speed: 70.700528 samples/sec	 accuracy=52.350446	 loss=2.016395	 lr=0.010000
Epoch[023] Batch [0749]/[3759]	Speed: 68.507695 samples/sec	 accuracy=52.366667	 loss=2.015535	 lr=0.010000
Epoch[023] Batch [0799]/[3759]	Speed: 70.163838 samples/sec	 accuracy=52.400391	 loss=2.013366	 lr=0.010000
Epoch[023] Batch [0849]/[3759]	Speed: 69.996928 samples/sec	 accuracy=52.450368	 loss=2.012963	 lr=0.010000
Epoch[023] Batch [0899]/[3759]	Speed: 69.149680 samples/sec	 accuracy=52.465278	 loss=2.010877	 lr=0.010000
Epoch[023] Batch [0949]/[3759]	Speed: 69.166755 samples/sec	 accuracy=52.493421	 loss=2.009422	 lr=0.010000
Epoch[023] Batch [0999]/[3759]	Speed: 70.189878 samples/sec	 accuracy=52.529688	 loss=2.007404	 lr=0.010000
Epoch[023] Batch [1049]/[3759]	Speed: 70.332575 samples/sec	 accuracy=52.507440	 loss=2.008241	 lr=0.010000
Epoch[023] Batch [1099]/[3759]	Speed: 70.614330 samples/sec	 accuracy=52.423295	 loss=2.011591	 lr=0.010000
Epoch[023] Batch [1149]/[3759]	Speed: 70.326318 samples/sec	 accuracy=52.421196	 loss=2.012280	 lr=0.010000
Epoch[023] Batch [1199]/[3759]	Speed: 68.085292 samples/sec	 accuracy=52.451823	 loss=2.012574	 lr=0.010000
Epoch[023] Batch [1249]/[3759]	Speed: 69.869375 samples/sec	 accuracy=52.407500	 loss=2.013882	 lr=0.010000
Epoch[023] Batch [1299]/[3759]	Speed: 68.705192 samples/sec	 accuracy=52.349760	 loss=2.015449	 lr=0.010000
Epoch[023] Batch [1349]/[3759]	Speed: 71.089393 samples/sec	 accuracy=52.305556	 loss=2.017633	 lr=0.010000
Epoch[023] Batch [1399]/[3759]	Speed: 70.739527 samples/sec	 accuracy=52.285714	 loss=2.017851	 lr=0.010000
Epoch[023] Batch [1449]/[3759]	Speed: 69.856844 samples/sec	 accuracy=52.301724	 loss=2.018586	 lr=0.010000
Epoch[023] Batch [1499]/[3759]	Speed: 70.050195 samples/sec	 accuracy=52.319792	 loss=2.017081	 lr=0.010000
Epoch[023] Batch [1549]/[3759]	Speed: 69.549347 samples/sec	 accuracy=52.311492	 loss=2.016868	 lr=0.010000
Epoch[023] Batch [1599]/[3759]	Speed: 69.333340 samples/sec	 accuracy=52.279297	 loss=2.016632	 lr=0.010000
Epoch[023] Batch [1649]/[3759]	Speed: 70.529109 samples/sec	 accuracy=52.251894	 loss=2.018726	 lr=0.010000
Epoch[023] Batch [1699]/[3759]	Speed: 69.462369 samples/sec	 accuracy=52.279412	 loss=2.018963	 lr=0.010000
Epoch[023] Batch [1749]/[3759]	Speed: 69.583247 samples/sec	 accuracy=52.283036	 loss=2.020718	 lr=0.010000
Epoch[023] Batch [1799]/[3759]	Speed: 67.894607 samples/sec	 accuracy=52.280382	 loss=2.021117	 lr=0.010000
Epoch[023] Batch [1849]/[3759]	Speed: 71.370976 samples/sec	 accuracy=52.285473	 loss=2.021179	 lr=0.010000
Epoch[023] Batch [1899]/[3759]	Speed: 69.528185 samples/sec	 accuracy=52.299342	 loss=2.021289	 lr=0.010000
Epoch[023] Batch [1949]/[3759]	Speed: 69.303421 samples/sec	 accuracy=52.337340	 loss=2.021377	 lr=0.010000
Epoch[023] Batch [1999]/[3759]	Speed: 68.909292 samples/sec	 accuracy=52.317187	 loss=2.023226	 lr=0.010000
Epoch[023] Batch [2049]/[3759]	Speed: 70.802496 samples/sec	 accuracy=52.302591	 loss=2.023733	 lr=0.010000
Epoch[023] Batch [2099]/[3759]	Speed: 70.171159 samples/sec	 accuracy=52.298363	 loss=2.023911	 lr=0.010000
Epoch[023] Batch [2149]/[3759]	Speed: 68.980072 samples/sec	 accuracy=52.293605	 loss=2.024280	 lr=0.010000
Epoch[023] Batch [2199]/[3759]	Speed: 69.508141 samples/sec	 accuracy=52.278409	 loss=2.024211	 lr=0.010000
Epoch[023] Batch [2249]/[3759]	Speed: 70.856266 samples/sec	 accuracy=52.251389	 loss=2.024821	 lr=0.010000
Epoch[023] Batch [2299]/[3759]	Speed: 69.301859 samples/sec	 accuracy=52.244565	 loss=2.025069	 lr=0.010000
Epoch[023] Batch [2349]/[3759]	Speed: 69.310500 samples/sec	 accuracy=52.246011	 loss=2.025529	 lr=0.010000
Epoch[023] Batch [2399]/[3759]	Speed: 67.177465 samples/sec	 accuracy=52.231771	 loss=2.026184	 lr=0.010000
Epoch[023] Batch [2449]/[3759]	Speed: 70.861858 samples/sec	 accuracy=52.202168	 loss=2.026860	 lr=0.010000
Epoch[023] Batch [2499]/[3759]	Speed: 69.572932 samples/sec	 accuracy=52.221875	 loss=2.026192	 lr=0.010000
Epoch[023] Batch [2549]/[3759]	Speed: 66.373920 samples/sec	 accuracy=52.221814	 loss=2.025597	 lr=0.010000
Epoch[023] Batch [2599]/[3759]	Speed: 72.603328 samples/sec	 accuracy=52.223558	 loss=2.025834	 lr=0.010000
Epoch[023] Batch [2649]/[3759]	Speed: 70.114780 samples/sec	 accuracy=52.250590	 loss=2.025838	 lr=0.010000
Epoch[023] Batch [2699]/[3759]	Speed: 70.388890 samples/sec	 accuracy=52.209491	 loss=2.027291	 lr=0.010000
Epoch[023] Batch [2749]/[3759]	Speed: 69.046737 samples/sec	 accuracy=52.226705	 loss=2.027407	 lr=0.010000
Epoch[023] Batch [2799]/[3759]	Speed: 70.178787 samples/sec	 accuracy=52.196987	 loss=2.028984	 lr=0.010000
Epoch[023] Batch [2849]/[3759]	Speed: 69.319245 samples/sec	 accuracy=52.194627	 loss=2.029640	 lr=0.010000
Epoch[023] Batch [2899]/[3759]	Speed: 70.273396 samples/sec	 accuracy=52.197737	 loss=2.029165	 lr=0.010000
Epoch[023] Batch [2949]/[3759]	Speed: 70.164691 samples/sec	 accuracy=52.204449	 loss=2.028596	 lr=0.010000
Epoch[023] Batch [2999]/[3759]	Speed: 68.447668 samples/sec	 accuracy=52.208854	 loss=2.027881	 lr=0.010000
Epoch[023] Batch [3049]/[3759]	Speed: 68.683123 samples/sec	 accuracy=52.211066	 loss=2.027811	 lr=0.010000
Epoch[023] Batch [3099]/[3759]	Speed: 68.352335 samples/sec	 accuracy=52.217238	 loss=2.027637	 lr=0.010000
Epoch[023] Batch [3149]/[3759]	Speed: 72.266617 samples/sec	 accuracy=52.203869	 loss=2.028368	 lr=0.010000
Epoch[023] Batch [3199]/[3759]	Speed: 69.961231 samples/sec	 accuracy=52.199219	 loss=2.028514	 lr=0.010000
Epoch[023] Batch [3249]/[3759]	Speed: 68.480849 samples/sec	 accuracy=52.210096	 loss=2.028259	 lr=0.010000
Epoch[023] Batch [3299]/[3759]	Speed: 70.059741 samples/sec	 accuracy=52.197917	 loss=2.028741	 lr=0.010000
Epoch[023] Batch [3349]/[3759]	Speed: 69.072856 samples/sec	 accuracy=52.196362	 loss=2.029462	 lr=0.010000
Epoch[023] Batch [3399]/[3759]	Speed: 69.559808 samples/sec	 accuracy=52.198529	 loss=2.029145	 lr=0.010000
Epoch[023] Batch [3449]/[3759]	Speed: 68.717513 samples/sec	 accuracy=52.176630	 loss=2.029613	 lr=0.010000
Epoch[023] Batch [3499]/[3759]	Speed: 70.121485 samples/sec	 accuracy=52.183482	 loss=2.030198	 lr=0.010000
Epoch[023] Batch [3549]/[3759]	Speed: 69.754924 samples/sec	 accuracy=52.188820	 loss=2.030553	 lr=0.010000
Epoch[023] Batch [3599]/[3759]	Speed: 68.819391 samples/sec	 accuracy=52.181424	 loss=2.030817	 lr=0.010000
Epoch[023] Batch [3649]/[3759]	Speed: 69.684435 samples/sec	 accuracy=52.158818	 loss=2.031440	 lr=0.010000
Epoch[023] Batch [3699]/[3759]	Speed: 70.312010 samples/sec	 accuracy=52.163429	 loss=2.031324	 lr=0.010000
Epoch[023] Batch [3749]/[3759]	Speed: 76.579714 samples/sec	 accuracy=52.145417	 loss=2.031032	 lr=0.010000
Batch [0049]/[0303]: acc-top1=52.906250 acc-top5=78.218750
Batch [0099]/[0303]: acc-top1=52.875000 acc-top5=77.968750
Batch [0149]/[0303]: acc-top1=53.729167 acc-top5=78.406250
Batch [0199]/[0303]: acc-top1=53.812500 acc-top5=78.507812
Batch [0249]/[0303]: acc-top1=53.850000 acc-top5=78.650000
Batch [0299]/[0303]: acc-top1=53.682292 acc-top5=78.484375
[Epoch 023] training: accuracy=52.148593	 loss=2.031284
[Epoch 023] speed: 69 samples/sec	time cost: 3760.910399
[Epoch 023] validation: acc-top1=53.676774 acc-top5=78.470503 loss=2.071454
Epoch[024] Batch [0049]/[3760]	Speed: 48.400569 samples/sec	 accuracy=52.937500	 loss=1.947750	 lr=0.010000
Epoch[024] Batch [0099]/[3760]	Speed: 67.346258 samples/sec	 accuracy=53.250000	 loss=1.954888	 lr=0.010000
Epoch[024] Batch [0149]/[3760]	Speed: 69.436941 samples/sec	 accuracy=53.364583	 loss=1.964151	 lr=0.010000
Epoch[024] Batch [0199]/[3760]	Speed: 68.759013 samples/sec	 accuracy=53.015625	 loss=1.982608	 lr=0.010000
Epoch[024] Batch [0249]/[3760]	Speed: 70.849103 samples/sec	 accuracy=53.000000	 loss=1.987337	 lr=0.010000
Epoch[024] Batch [0299]/[3760]	Speed: 67.751286 samples/sec	 accuracy=53.005208	 loss=1.991647	 lr=0.010000
Epoch[024] Batch [0349]/[3760]	Speed: 70.747952 samples/sec	 accuracy=52.991071	 loss=1.990805	 lr=0.010000
Epoch[024] Batch [0399]/[3760]	Speed: 67.911168 samples/sec	 accuracy=52.832031	 loss=1.993842	 lr=0.010000
Epoch[024] Batch [0449]/[3760]	Speed: 70.792148 samples/sec	 accuracy=52.888889	 loss=1.991366	 lr=0.010000
Epoch[024] Batch [0499]/[3760]	Speed: 71.399659 samples/sec	 accuracy=52.793750	 loss=1.992014	 lr=0.010000
Epoch[024] Batch [0549]/[3760]	Speed: 68.382052 samples/sec	 accuracy=52.681818	 loss=1.999953	 lr=0.010000
Epoch[024] Batch [0599]/[3760]	Speed: 68.633993 samples/sec	 accuracy=52.609375	 loss=2.003758	 lr=0.010000
Epoch[024] Batch [0649]/[3760]	Speed: 69.713926 samples/sec	 accuracy=52.620192	 loss=2.005412	 lr=0.010000
Epoch[024] Batch [0699]/[3760]	Speed: 70.296544 samples/sec	 accuracy=52.676339	 loss=2.005155	 lr=0.010000
Epoch[024] Batch [0749]/[3760]	Speed: 69.376227 samples/sec	 accuracy=52.616667	 loss=2.004533	 lr=0.010000
Epoch[024] Batch [0799]/[3760]	Speed: 67.807787 samples/sec	 accuracy=52.531250	 loss=2.009222	 lr=0.010000
Epoch[024] Batch [0849]/[3760]	Speed: 69.684085 samples/sec	 accuracy=52.509191	 loss=2.006706	 lr=0.010000
Epoch[024] Batch [0899]/[3760]	Speed: 71.427036 samples/sec	 accuracy=52.543403	 loss=2.004353	 lr=0.010000
Epoch[024] Batch [0949]/[3760]	Speed: 69.371620 samples/sec	 accuracy=52.570724	 loss=2.003026	 lr=0.010000
Epoch[024] Batch [0999]/[3760]	Speed: 70.256358 samples/sec	 accuracy=52.614063	 loss=2.003574	 lr=0.010000
Epoch[024] Batch [1049]/[3760]	Speed: 68.339633 samples/sec	 accuracy=52.620536	 loss=2.004624	 lr=0.010000
Epoch[024] Batch [1099]/[3760]	Speed: 70.757615 samples/sec	 accuracy=52.568182	 loss=2.006680	 lr=0.010000
Epoch[024] Batch [1149]/[3760]	Speed: 68.036847 samples/sec	 accuracy=52.601902	 loss=2.004620	 lr=0.010000
Epoch[024] Batch [1199]/[3760]	Speed: 70.953895 samples/sec	 accuracy=52.621094	 loss=2.005070	 lr=0.010000
Epoch[024] Batch [1249]/[3760]	Speed: 67.791929 samples/sec	 accuracy=52.647500	 loss=2.004679	 lr=0.010000
Epoch[024] Batch [1299]/[3760]	Speed: 71.179747 samples/sec	 accuracy=52.606971	 loss=2.006566	 lr=0.010000
Epoch[024] Batch [1349]/[3760]	Speed: 68.698910 samples/sec	 accuracy=52.615741	 loss=2.006531	 lr=0.010000
Epoch[024] Batch [1399]/[3760]	Speed: 69.937626 samples/sec	 accuracy=52.604911	 loss=2.006247	 lr=0.010000
Epoch[024] Batch [1449]/[3760]	Speed: 69.341924 samples/sec	 accuracy=52.529095	 loss=2.007843	 lr=0.010000
Epoch[024] Batch [1499]/[3760]	Speed: 69.961526 samples/sec	 accuracy=52.553125	 loss=2.007816	 lr=0.010000
Epoch[024] Batch [1549]/[3760]	Speed: 69.189144 samples/sec	 accuracy=52.553427	 loss=2.009312	 lr=0.010000
Epoch[024] Batch [1599]/[3760]	Speed: 69.859434 samples/sec	 accuracy=52.555664	 loss=2.009341	 lr=0.010000
Epoch[024] Batch [1649]/[3760]	Speed: 69.566275 samples/sec	 accuracy=52.547348	 loss=2.009269	 lr=0.010000
Epoch[024] Batch [1699]/[3760]	Speed: 65.650736 samples/sec	 accuracy=52.525735	 loss=2.010445	 lr=0.010000
Epoch[024] Batch [1749]/[3760]	Speed: 72.413500 samples/sec	 accuracy=52.502679	 loss=2.011043	 lr=0.010000
Epoch[024] Batch [1799]/[3760]	Speed: 66.954812 samples/sec	 accuracy=52.467014	 loss=2.012471	 lr=0.010000
Epoch[024] Batch [1849]/[3760]	Speed: 71.178238 samples/sec	 accuracy=52.516892	 loss=2.011267	 lr=0.010000
Epoch[024] Batch [1899]/[3760]	Speed: 70.529231 samples/sec	 accuracy=52.466283	 loss=2.012827	 lr=0.010000
Epoch[024] Batch [1949]/[3760]	Speed: 69.017024 samples/sec	 accuracy=52.493590	 loss=2.011201	 lr=0.010000
Epoch[024] Batch [1999]/[3760]	Speed: 69.107110 samples/sec	 accuracy=52.464844	 loss=2.011388	 lr=0.010000
Epoch[024] Batch [2049]/[3760]	Speed: 69.652432 samples/sec	 accuracy=52.495427	 loss=2.010030	 lr=0.010000
Epoch[024] Batch [2099]/[3760]	Speed: 69.056969 samples/sec	 accuracy=52.470982	 loss=2.010624	 lr=0.010000
Epoch[024] Batch [2149]/[3760]	Speed: 70.124485 samples/sec	 accuracy=52.457849	 loss=2.010770	 lr=0.010000
Epoch[024] Batch [2199]/[3760]	Speed: 65.796773 samples/sec	 accuracy=52.465199	 loss=2.009744	 lr=0.010000
Epoch[024] Batch [2249]/[3760]	Speed: 71.088282 samples/sec	 accuracy=52.476389	 loss=2.009648	 lr=0.010000
Epoch[024] Batch [2299]/[3760]	Speed: 70.129312 samples/sec	 accuracy=52.463315	 loss=2.010529	 lr=0.010000
Epoch[024] Batch [2349]/[3760]	Speed: 70.130982 samples/sec	 accuracy=52.462766	 loss=2.010638	 lr=0.010000
Epoch[024] Batch [2399]/[3760]	Speed: 70.374916 samples/sec	 accuracy=52.470052	 loss=2.009898	 lr=0.010000
Epoch[024] Batch [2449]/[3760]	Speed: 68.610719 samples/sec	 accuracy=52.454719	 loss=2.010652	 lr=0.010000
Epoch[024] Batch [2499]/[3760]	Speed: 70.275622 samples/sec	 accuracy=52.451875	 loss=2.010243	 lr=0.010000
Epoch[024] Batch [2549]/[3760]	Speed: 67.967116 samples/sec	 accuracy=52.449755	 loss=2.009924	 lr=0.010000
Epoch[024] Batch [2599]/[3760]	Speed: 69.188956 samples/sec	 accuracy=52.412861	 loss=2.012145	 lr=0.010000
Epoch[024] Batch [2649]/[3760]	Speed: 68.040717 samples/sec	 accuracy=52.422759	 loss=2.012073	 lr=0.010000
Epoch[024] Batch [2699]/[3760]	Speed: 69.987001 samples/sec	 accuracy=52.442708	 loss=2.012217	 lr=0.010000
Epoch[024] Batch [2749]/[3760]	Speed: 71.291276 samples/sec	 accuracy=52.432386	 loss=2.012381	 lr=0.010000
Epoch[024] Batch [2799]/[3760]	Speed: 69.946313 samples/sec	 accuracy=52.400670	 loss=2.013246	 lr=0.010000
Epoch[024] Batch [2849]/[3760]	Speed: 68.694866 samples/sec	 accuracy=52.425987	 loss=2.012096	 lr=0.010000
Epoch[024] Batch [2899]/[3760]	Speed: 70.197455 samples/sec	 accuracy=52.435884	 loss=2.012072	 lr=0.010000
Epoch[024] Batch [2949]/[3760]	Speed: 70.797765 samples/sec	 accuracy=52.441208	 loss=2.012414	 lr=0.010000
Epoch[024] Batch [2999]/[3760]	Speed: 69.440760 samples/sec	 accuracy=52.444792	 loss=2.012567	 lr=0.010000
Epoch[024] Batch [3049]/[3760]	Speed: 69.149377 samples/sec	 accuracy=52.473873	 loss=2.011233	 lr=0.010000
Epoch[024] Batch [3099]/[3760]	Speed: 68.798479 samples/sec	 accuracy=52.461190	 loss=2.012487	 lr=0.010000
Epoch[024] Batch [3149]/[3760]	Speed: 69.207342 samples/sec	 accuracy=52.435020	 loss=2.013011	 lr=0.010000
Epoch[024] Batch [3199]/[3760]	Speed: 70.309961 samples/sec	 accuracy=52.424805	 loss=2.013638	 lr=0.010000
Epoch[024] Batch [3249]/[3760]	Speed: 69.540218 samples/sec	 accuracy=52.437981	 loss=2.013010	 lr=0.010000
Epoch[024] Batch [3299]/[3760]	Speed: 70.814764 samples/sec	 accuracy=52.456439	 loss=2.012182	 lr=0.010000
Epoch[024] Batch [3349]/[3760]	Speed: 67.809788 samples/sec	 accuracy=52.455224	 loss=2.012383	 lr=0.010000
Epoch[024] Batch [3399]/[3760]	Speed: 70.249226 samples/sec	 accuracy=52.431066	 loss=2.013324	 lr=0.010000
Epoch[024] Batch [3449]/[3760]	Speed: 70.708187 samples/sec	 accuracy=52.417572	 loss=2.013839	 lr=0.010000
Epoch[024] Batch [3499]/[3760]	Speed: 68.485486 samples/sec	 accuracy=52.383036	 loss=2.014988	 lr=0.010000
Epoch[024] Batch [3549]/[3760]	Speed: 70.699046 samples/sec	 accuracy=52.382923	 loss=2.015331	 lr=0.010000
Epoch[024] Batch [3599]/[3760]	Speed: 68.900744 samples/sec	 accuracy=52.356771	 loss=2.015772	 lr=0.010000
Epoch[024] Batch [3649]/[3760]	Speed: 70.374415 samples/sec	 accuracy=52.385274	 loss=2.014988	 lr=0.010000
Epoch[024] Batch [3699]/[3760]	Speed: 70.026588 samples/sec	 accuracy=52.379223	 loss=2.015476	 lr=0.010000
Epoch[024] Batch [3749]/[3760]	Speed: 76.902956 samples/sec	 accuracy=52.392083	 loss=2.015374	 lr=0.010000
Batch [0049]/[0303]: acc-top1=54.187500 acc-top5=77.500000
Batch [0099]/[0303]: acc-top1=53.093750 acc-top5=77.687500
Batch [0149]/[0303]: acc-top1=53.479167 acc-top5=78.322917
Batch [0199]/[0303]: acc-top1=53.703125 acc-top5=78.507812
Batch [0249]/[0303]: acc-top1=53.725000 acc-top5=78.656250
Batch [0299]/[0303]: acc-top1=53.588542 acc-top5=78.265625
[Epoch 024] training: accuracy=52.397773	 loss=2.015070
[Epoch 024] speed: 69 samples/sec	time cost: 3769.472700
[Epoch 024] validation: acc-top1=53.547855 acc-top5=78.233292 loss=2.084351
Epoch[025] Batch [0049]/[3760]	Speed: 49.028566 samples/sec	 accuracy=54.125000	 loss=1.913322	 lr=0.010000
Epoch[025] Batch [0099]/[3760]	Speed: 66.488194 samples/sec	 accuracy=54.062500	 loss=1.925994	 lr=0.010000
Epoch[025] Batch [0149]/[3760]	Speed: 69.630917 samples/sec	 accuracy=52.989583	 loss=1.962032	 lr=0.010000
Epoch[025] Batch [0199]/[3760]	Speed: 70.290101 samples/sec	 accuracy=53.023438	 loss=1.962740	 lr=0.010000
Epoch[025] Batch [0249]/[3760]	Speed: 68.053162 samples/sec	 accuracy=53.143750	 loss=1.958973	 lr=0.010000
Epoch[025] Batch [0299]/[3760]	Speed: 71.536619 samples/sec	 accuracy=53.010417	 loss=1.965715	 lr=0.010000
Epoch[025] Batch [0349]/[3760]	Speed: 69.905411 samples/sec	 accuracy=53.022321	 loss=1.965396	 lr=0.010000
Epoch[025] Batch [0399]/[3760]	Speed: 70.239327 samples/sec	 accuracy=52.992188	 loss=1.966256	 lr=0.010000
Epoch[025] Batch [0449]/[3760]	Speed: 68.532832 samples/sec	 accuracy=52.944444	 loss=1.972285	 lr=0.010000
Epoch[025] Batch [0499]/[3760]	Speed: 70.611766 samples/sec	 accuracy=52.781250	 loss=1.978323	 lr=0.010000
Epoch[025] Batch [0549]/[3760]	Speed: 67.204842 samples/sec	 accuracy=52.747159	 loss=1.980687	 lr=0.010000
Epoch[025] Batch [0599]/[3760]	Speed: 70.642607 samples/sec	 accuracy=52.781250	 loss=1.984085	 lr=0.010000
Epoch[025] Batch [0649]/[3760]	Speed: 69.500423 samples/sec	 accuracy=52.802885	 loss=1.985414	 lr=0.010000
Epoch[025] Batch [0699]/[3760]	Speed: 70.019570 samples/sec	 accuracy=52.792411	 loss=1.985435	 lr=0.010000
Epoch[025] Batch [0749]/[3760]	Speed: 68.639889 samples/sec	 accuracy=52.758333	 loss=1.986530	 lr=0.010000
Epoch[025] Batch [0799]/[3760]	Speed: 70.893411 samples/sec	 accuracy=52.822266	 loss=1.984739	 lr=0.010000
Epoch[025] Batch [0849]/[3760]	Speed: 69.456286 samples/sec	 accuracy=52.994485	 loss=1.979125	 lr=0.010000
Epoch[025] Batch [0899]/[3760]	Speed: 70.239178 samples/sec	 accuracy=52.965278	 loss=1.980899	 lr=0.010000
Epoch[025] Batch [0949]/[3760]	Speed: 68.587056 samples/sec	 accuracy=52.921053	 loss=1.982655	 lr=0.010000
Epoch[025] Batch [0999]/[3760]	Speed: 71.261820 samples/sec	 accuracy=52.910938	 loss=1.983888	 lr=0.010000
Epoch[025] Batch [1049]/[3760]	Speed: 69.069411 samples/sec	 accuracy=52.819940	 loss=1.988321	 lr=0.010000
Epoch[025] Batch [1099]/[3760]	Speed: 70.152329 samples/sec	 accuracy=52.785511	 loss=1.990478	 lr=0.010000
Epoch[025] Batch [1149]/[3760]	Speed: 70.426702 samples/sec	 accuracy=52.762228	 loss=1.991822	 lr=0.010000
Epoch[025] Batch [1199]/[3760]	Speed: 68.212223 samples/sec	 accuracy=52.695312	 loss=1.996737	 lr=0.010000
Epoch[025] Batch [1249]/[3760]	Speed: 70.438606 samples/sec	 accuracy=52.685000	 loss=1.997260	 lr=0.010000
Epoch[025] Batch [1299]/[3760]	Speed: 69.015253 samples/sec	 accuracy=52.686298	 loss=1.997549	 lr=0.010000
Epoch[025] Batch [1349]/[3760]	Speed: 70.176914 samples/sec	 accuracy=52.674769	 loss=1.999090	 lr=0.010000
Epoch[025] Batch [1399]/[3760]	Speed: 69.501800 samples/sec	 accuracy=52.698661	 loss=1.998139	 lr=0.010000
Epoch[025] Batch [1449]/[3760]	Speed: 70.362824 samples/sec	 accuracy=52.667026	 loss=1.998314	 lr=0.010000
Epoch[025] Batch [1499]/[3760]	Speed: 69.957418 samples/sec	 accuracy=52.712500	 loss=1.997418	 lr=0.010000
Epoch[025] Batch [1549]/[3760]	Speed: 69.171997 samples/sec	 accuracy=52.745968	 loss=1.997101	 lr=0.010000
Epoch[025] Batch [1599]/[3760]	Speed: 69.827399 samples/sec	 accuracy=52.768555	 loss=1.996578	 lr=0.010000
Epoch[025] Batch [1649]/[3760]	Speed: 70.508696 samples/sec	 accuracy=52.749053	 loss=1.997120	 lr=0.010000
Epoch[025] Batch [1699]/[3760]	Speed: 70.228440 samples/sec	 accuracy=52.707721	 loss=2.000335	 lr=0.010000
Epoch[025] Batch [1749]/[3760]	Speed: 69.223425 samples/sec	 accuracy=52.732143	 loss=2.000013	 lr=0.010000
Epoch[025] Batch [1799]/[3760]	Speed: 70.725549 samples/sec	 accuracy=52.707465	 loss=2.001239	 lr=0.010000
Epoch[025] Batch [1849]/[3760]	Speed: 69.486079 samples/sec	 accuracy=52.688345	 loss=2.001803	 lr=0.010000
Epoch[025] Batch [1899]/[3760]	Speed: 68.892629 samples/sec	 accuracy=52.654605	 loss=2.003071	 lr=0.010000
Epoch[025] Batch [1949]/[3760]	Speed: 70.689136 samples/sec	 accuracy=52.654647	 loss=2.003517	 lr=0.010000
Epoch[025] Batch [1999]/[3760]	Speed: 69.548049 samples/sec	 accuracy=52.676563	 loss=2.003408	 lr=0.010000
Epoch[025] Batch [2049]/[3760]	Speed: 69.251130 samples/sec	 accuracy=52.675305	 loss=2.003785	 lr=0.010000
Epoch[025] Batch [2099]/[3760]	Speed: 68.550993 samples/sec	 accuracy=52.677827	 loss=2.004363	 lr=0.010000
Epoch[025] Batch [2149]/[3760]	Speed: 69.938928 samples/sec	 accuracy=52.685320	 loss=2.004000	 lr=0.010000
Epoch[025] Batch [2199]/[3760]	Speed: 70.206630 samples/sec	 accuracy=52.692472	 loss=2.003505	 lr=0.010000
Epoch[025] Batch [2249]/[3760]	Speed: 68.272990 samples/sec	 accuracy=52.688194	 loss=2.003777	 lr=0.010000
Epoch[025] Batch [2299]/[3760]	Speed: 70.626289 samples/sec	 accuracy=52.695652	 loss=2.003517	 lr=0.010000
Epoch[025] Batch [2349]/[3760]	Speed: 69.232439 samples/sec	 accuracy=52.695479	 loss=2.003301	 lr=0.010000
Epoch[025] Batch [2399]/[3760]	Speed: 70.401554 samples/sec	 accuracy=52.710938	 loss=2.002894	 lr=0.010000
Epoch[025] Batch [2449]/[3760]	Speed: 70.216302 samples/sec	 accuracy=52.754464	 loss=2.001240	 lr=0.010000
Epoch[025] Batch [2499]/[3760]	Speed: 68.813939 samples/sec	 accuracy=52.787500	 loss=1.999957	 lr=0.010000
Epoch[025] Batch [2549]/[3760]	Speed: 70.089793 samples/sec	 accuracy=52.780025	 loss=1.999622	 lr=0.010000
Epoch[025] Batch [2599]/[3760]	Speed: 70.367356 samples/sec	 accuracy=52.745192	 loss=2.000759	 lr=0.010000
Epoch[025] Batch [2649]/[3760]	Speed: 68.464597 samples/sec	 accuracy=52.751769	 loss=2.000915	 lr=0.010000
Epoch[025] Batch [2699]/[3760]	Speed: 69.943102 samples/sec	 accuracy=52.758681	 loss=2.000980	 lr=0.010000
Epoch[025] Batch [2749]/[3760]	Speed: 71.012343 samples/sec	 accuracy=52.722727	 loss=2.002393	 lr=0.010000
Epoch[025] Batch [2799]/[3760]	Speed: 69.480227 samples/sec	 accuracy=52.716518	 loss=2.001781	 lr=0.010000
Epoch[025] Batch [2849]/[3760]	Speed: 70.172644 samples/sec	 accuracy=52.718750	 loss=2.001915	 lr=0.010000
Epoch[025] Batch [2899]/[3760]	Speed: 69.458069 samples/sec	 accuracy=52.698276	 loss=2.003438	 lr=0.010000
Epoch[025] Batch [2949]/[3760]	Speed: 69.220322 samples/sec	 accuracy=52.675318	 loss=2.004660	 lr=0.010000
Epoch[025] Batch [2999]/[3760]	Speed: 68.015984 samples/sec	 accuracy=52.681250	 loss=2.004542	 lr=0.010000
Epoch[025] Batch [3049]/[3760]	Speed: 70.906740 samples/sec	 accuracy=52.689037	 loss=2.004833	 lr=0.010000
Epoch[025] Batch [3099]/[3760]	Speed: 68.513229 samples/sec	 accuracy=52.692036	 loss=2.005489	 lr=0.010000
Epoch[025] Batch [3149]/[3760]	Speed: 70.462071 samples/sec	 accuracy=52.679067	 loss=2.005242	 lr=0.010000
Epoch[025] Batch [3199]/[3760]	Speed: 69.455127 samples/sec	 accuracy=52.671875	 loss=2.005854	 lr=0.010000
Epoch[025] Batch [3249]/[3760]	Speed: 69.545159 samples/sec	 accuracy=52.670192	 loss=2.006413	 lr=0.010000
Epoch[025] Batch [3299]/[3760]	Speed: 69.889346 samples/sec	 accuracy=52.662879	 loss=2.007151	 lr=0.010000
Epoch[025] Batch [3349]/[3760]	Speed: 70.643036 samples/sec	 accuracy=52.683769	 loss=2.006473	 lr=0.010000
Epoch[025] Batch [3399]/[3760]	Speed: 69.190064 samples/sec	 accuracy=52.658548	 loss=2.006925	 lr=0.010000
Epoch[025] Batch [3449]/[3760]	Speed: 70.828152 samples/sec	 accuracy=52.660326	 loss=2.006891	 lr=0.010000
Epoch[025] Batch [3499]/[3760]	Speed: 67.952517 samples/sec	 accuracy=52.675446	 loss=2.006595	 lr=0.010000
Epoch[025] Batch [3549]/[3760]	Speed: 70.521341 samples/sec	 accuracy=52.673856	 loss=2.006771	 lr=0.010000
Epoch[025] Batch [3599]/[3760]	Speed: 70.950942 samples/sec	 accuracy=52.675347	 loss=2.006604	 lr=0.010000
Epoch[025] Batch [3649]/[3760]	Speed: 68.895060 samples/sec	 accuracy=52.672945	 loss=2.006393	 lr=0.010000
Epoch[025] Batch [3699]/[3760]	Speed: 68.979884 samples/sec	 accuracy=52.672720	 loss=2.006818	 lr=0.010000
Epoch[025] Batch [3749]/[3760]	Speed: 78.651838 samples/sec	 accuracy=52.674167	 loss=2.006928	 lr=0.010000
Batch [0049]/[0303]: acc-top1=52.156250 acc-top5=77.093750
Batch [0099]/[0303]: acc-top1=51.500000 acc-top5=76.734375
Batch [0149]/[0303]: acc-top1=51.979167 acc-top5=77.531250
Batch [0199]/[0303]: acc-top1=52.226562 acc-top5=77.656250
Batch [0249]/[0303]: acc-top1=52.412500 acc-top5=77.775000
Batch [0299]/[0303]: acc-top1=52.166667 acc-top5=77.583333
[Epoch 025] training: accuracy=52.690741	 loss=2.006207
[Epoch 025] speed: 69 samples/sec	time cost: 3759.373072
[Epoch 025] validation: acc-top1=52.129744 acc-top5=77.557756 loss=2.177764
Epoch[026] Batch [0049]/[3759]	Speed: 48.007163 samples/sec	 accuracy=54.875000	 loss=1.958715	 lr=0.010000
Epoch[026] Batch [0099]/[3759]	Speed: 68.542039 samples/sec	 accuracy=54.218750	 loss=1.979236	 lr=0.010000
Epoch[026] Batch [0149]/[3759]	Speed: 68.479268 samples/sec	 accuracy=54.052083	 loss=1.978937	 lr=0.010000
Epoch[026] Batch [0199]/[3759]	Speed: 70.479758 samples/sec	 accuracy=53.968750	 loss=1.972162	 lr=0.010000
Epoch[026] Batch [0249]/[3759]	Speed: 70.865906 samples/sec	 accuracy=54.043750	 loss=1.962668	 lr=0.010000
Epoch[026] Batch [0299]/[3759]	Speed: 68.415336 samples/sec	 accuracy=53.708333	 loss=1.969234	 lr=0.010000
Epoch[026] Batch [0349]/[3759]	Speed: 69.423074 samples/sec	 accuracy=53.633929	 loss=1.970948	 lr=0.010000
Epoch[026] Batch [0399]/[3759]	Speed: 70.994623 samples/sec	 accuracy=53.550781	 loss=1.970872	 lr=0.010000
Epoch[026] Batch [0449]/[3759]	Speed: 70.233998 samples/sec	 accuracy=53.562500	 loss=1.969204	 lr=0.010000
Epoch[026] Batch [0499]/[3759]	Speed: 70.217628 samples/sec	 accuracy=53.728125	 loss=1.969555	 lr=0.010000
Epoch[026] Batch [0549]/[3759]	Speed: 68.606932 samples/sec	 accuracy=53.784091	 loss=1.966530	 lr=0.010000
Epoch[026] Batch [0599]/[3759]	Speed: 70.253905 samples/sec	 accuracy=53.776042	 loss=1.968957	 lr=0.010000
Epoch[026] Batch [0649]/[3759]	Speed: 69.926490 samples/sec	 accuracy=53.838942	 loss=1.967474	 lr=0.010000
Epoch[026] Batch [0699]/[3759]	Speed: 69.583675 samples/sec	 accuracy=53.754464	 loss=1.965467	 lr=0.010000
Epoch[026] Batch [0749]/[3759]	Speed: 70.573205 samples/sec	 accuracy=53.747917	 loss=1.965661	 lr=0.010000
Epoch[026] Batch [0799]/[3759]	Speed: 69.751207 samples/sec	 accuracy=53.625000	 loss=1.968014	 lr=0.010000
Epoch[026] Batch [0849]/[3759]	Speed: 69.633829 samples/sec	 accuracy=53.547794	 loss=1.967760	 lr=0.010000
Epoch[026] Batch [0899]/[3759]	Speed: 69.817031 samples/sec	 accuracy=53.506944	 loss=1.967757	 lr=0.010000
Epoch[026] Batch [0949]/[3759]	Speed: 70.127795 samples/sec	 accuracy=53.486842	 loss=1.968898	 lr=0.010000
Epoch[026] Batch [0999]/[3759]	Speed: 69.474425 samples/sec	 accuracy=53.443750	 loss=1.968342	 lr=0.010000
Epoch[026] Batch [1049]/[3759]	Speed: 69.842944 samples/sec	 accuracy=53.430060	 loss=1.969669	 lr=0.010000
Epoch[026] Batch [1099]/[3759]	Speed: 70.208051 samples/sec	 accuracy=53.392045	 loss=1.970786	 lr=0.010000
Epoch[026] Batch [1149]/[3759]	Speed: 69.439733 samples/sec	 accuracy=53.379076	 loss=1.969949	 lr=0.010000
Epoch[026] Batch [1199]/[3759]	Speed: 69.209758 samples/sec	 accuracy=53.334635	 loss=1.972039	 lr=0.010000
Epoch[026] Batch [1249]/[3759]	Speed: 70.130053 samples/sec	 accuracy=53.290000	 loss=1.972720	 lr=0.010000
Epoch[026] Batch [1299]/[3759]	Speed: 71.649245 samples/sec	 accuracy=53.247596	 loss=1.975739	 lr=0.010000
Epoch[026] Batch [1349]/[3759]	Speed: 69.266424 samples/sec	 accuracy=53.212963	 loss=1.977440	 lr=0.010000
Epoch[026] Batch [1399]/[3759]	Speed: 69.377073 samples/sec	 accuracy=53.209821	 loss=1.977498	 lr=0.010000
Epoch[026] Batch [1449]/[3759]	Speed: 69.923639 samples/sec	 accuracy=53.185345	 loss=1.978391	 lr=0.010000
Epoch[026] Batch [1499]/[3759]	Speed: 70.099342 samples/sec	 accuracy=53.152083	 loss=1.979747	 lr=0.010000
Epoch[026] Batch [1549]/[3759]	Speed: 69.086359 samples/sec	 accuracy=53.141129	 loss=1.981295	 lr=0.010000
Epoch[026] Batch [1599]/[3759]	Speed: 69.875788 samples/sec	 accuracy=53.110352	 loss=1.982817	 lr=0.010000
Epoch[026] Batch [1649]/[3759]	Speed: 71.079972 samples/sec	 accuracy=53.075758	 loss=1.983801	 lr=0.010000
Epoch[026] Batch [1699]/[3759]	Speed: 69.254238 samples/sec	 accuracy=53.098346	 loss=1.981949	 lr=0.010000
Epoch[026] Batch [1749]/[3759]	Speed: 70.237146 samples/sec	 accuracy=53.132143	 loss=1.979443	 lr=0.010000
Epoch[026] Batch [1799]/[3759]	Speed: 68.765806 samples/sec	 accuracy=53.089410	 loss=1.981036	 lr=0.010000
Epoch[026] Batch [1849]/[3759]	Speed: 70.505996 samples/sec	 accuracy=53.080236	 loss=1.981792	 lr=0.010000
Epoch[026] Batch [1899]/[3759]	Speed: 68.514287 samples/sec	 accuracy=53.075658	 loss=1.981457	 lr=0.010000
Epoch[026] Batch [1949]/[3759]	Speed: 69.591236 samples/sec	 accuracy=53.048878	 loss=1.982334	 lr=0.010000
Epoch[026] Batch [1999]/[3759]	Speed: 69.046403 samples/sec	 accuracy=53.050000	 loss=1.983572	 lr=0.010000
Epoch[026] Batch [2049]/[3759]	Speed: 69.276550 samples/sec	 accuracy=52.991616	 loss=1.986658	 lr=0.010000
Epoch[026] Batch [2099]/[3759]	Speed: 71.564921 samples/sec	 accuracy=52.971726	 loss=1.988150	 lr=0.010000
Epoch[026] Batch [2149]/[3759]	Speed: 68.949480 samples/sec	 accuracy=52.920785	 loss=1.990794	 lr=0.010000
Epoch[026] Batch [2199]/[3759]	Speed: 70.440275 samples/sec	 accuracy=52.904830	 loss=1.991346	 lr=0.010000
Epoch[026] Batch [2249]/[3759]	Speed: 70.963689 samples/sec	 accuracy=52.897917	 loss=1.991014	 lr=0.010000
Epoch[026] Batch [2299]/[3759]	Speed: 67.846877 samples/sec	 accuracy=52.909647	 loss=1.990254	 lr=0.010000
Epoch[026] Batch [2349]/[3759]	Speed: 69.750267 samples/sec	 accuracy=52.920213	 loss=1.989739	 lr=0.010000
Epoch[026] Batch [2399]/[3759]	Speed: 70.134853 samples/sec	 accuracy=52.885417	 loss=1.991822	 lr=0.010000
Epoch[026] Batch [2449]/[3759]	Speed: 70.744941 samples/sec	 accuracy=52.882653	 loss=1.991632	 lr=0.010000
Epoch[026] Batch [2499]/[3759]	Speed: 68.344158 samples/sec	 accuracy=52.865625	 loss=1.992225	 lr=0.010000
Epoch[026] Batch [2549]/[3759]	Speed: 70.133597 samples/sec	 accuracy=52.888480	 loss=1.991517	 lr=0.010000
Epoch[026] Batch [2599]/[3759]	Speed: 68.406708 samples/sec	 accuracy=52.882212	 loss=1.991850	 lr=0.010000
Epoch[026] Batch [2649]/[3759]	Speed: 70.975570 samples/sec	 accuracy=52.863208	 loss=1.993211	 lr=0.010000
Epoch[026] Batch [2699]/[3759]	Speed: 69.017302 samples/sec	 accuracy=52.839120	 loss=1.994219	 lr=0.010000
Epoch[026] Batch [2749]/[3759]	Speed: 70.209906 samples/sec	 accuracy=52.842045	 loss=1.994768	 lr=0.010000
Epoch[026] Batch [2799]/[3759]	Speed: 69.462664 samples/sec	 accuracy=52.860491	 loss=1.993885	 lr=0.010000
Epoch[026] Batch [2849]/[3759]	Speed: 69.461153 samples/sec	 accuracy=52.876645	 loss=1.992884	 lr=0.010000
Epoch[026] Batch [2899]/[3759]	Speed: 69.895516 samples/sec	 accuracy=52.873384	 loss=1.993021	 lr=0.010000
Epoch[026] Batch [2949]/[3759]	Speed: 69.127333 samples/sec	 accuracy=52.868114	 loss=1.993473	 lr=0.010000
Epoch[026] Batch [2999]/[3759]	Speed: 70.594842 samples/sec	 accuracy=52.835938	 loss=1.994956	 lr=0.010000
Epoch[026] Batch [3049]/[3759]	Speed: 68.787260 samples/sec	 accuracy=52.807377	 loss=1.996148	 lr=0.010000
Epoch[026] Batch [3099]/[3759]	Speed: 68.824539 samples/sec	 accuracy=52.775706	 loss=1.997412	 lr=0.010000
Epoch[026] Batch [3149]/[3759]	Speed: 69.291079 samples/sec	 accuracy=52.788194	 loss=1.996990	 lr=0.010000
Epoch[026] Batch [3199]/[3759]	Speed: 70.062014 samples/sec	 accuracy=52.770996	 loss=1.997528	 lr=0.010000
Epoch[026] Batch [3249]/[3759]	Speed: 67.312200 samples/sec	 accuracy=52.800000	 loss=1.997786	 lr=0.010000
Epoch[026] Batch [3299]/[3759]	Speed: 71.450819 samples/sec	 accuracy=52.811080	 loss=1.997681	 lr=0.010000
Epoch[026] Batch [3349]/[3759]	Speed: 69.076652 samples/sec	 accuracy=52.808302	 loss=1.998463	 lr=0.010000
Epoch[026] Batch [3399]/[3759]	Speed: 69.858612 samples/sec	 accuracy=52.813419	 loss=1.998203	 lr=0.010000
Epoch[026] Batch [3449]/[3759]	Speed: 70.551200 samples/sec	 accuracy=52.805707	 loss=1.998291	 lr=0.010000
Epoch[026] Batch [3499]/[3759]	Speed: 69.128217 samples/sec	 accuracy=52.798214	 loss=1.998813	 lr=0.010000
Epoch[026] Batch [3549]/[3759]	Speed: 69.694940 samples/sec	 accuracy=52.776849	 loss=1.999875	 lr=0.010000
Epoch[026] Batch [3599]/[3759]	Speed: 67.237674 samples/sec	 accuracy=52.759549	 loss=2.000722	 lr=0.010000
Epoch[026] Batch [3649]/[3759]	Speed: 71.900301 samples/sec	 accuracy=52.738870	 loss=2.001599	 lr=0.010000
Epoch[026] Batch [3699]/[3759]	Speed: 68.426916 samples/sec	 accuracy=52.755912	 loss=2.000823	 lr=0.010000
Epoch[026] Batch [3749]/[3759]	Speed: 79.654472 samples/sec	 accuracy=52.746667	 loss=2.001601	 lr=0.010000
Batch [0049]/[0303]: acc-top1=54.625000 acc-top5=78.125000
Batch [0099]/[0303]: acc-top1=54.187500 acc-top5=78.421875
Batch [0149]/[0303]: acc-top1=54.385417 acc-top5=78.822917
Batch [0199]/[0303]: acc-top1=54.578125 acc-top5=78.968750
Batch [0249]/[0303]: acc-top1=54.656250 acc-top5=79.118750
Batch [0299]/[0303]: acc-top1=54.473958 acc-top5=78.848958
[Epoch 026] training: accuracy=52.747988	 loss=2.001776
[Epoch 026] speed: 69 samples/sec	time cost: 3761.471425
[Epoch 026] validation: acc-top1=54.429662 acc-top5=78.795380 loss=2.028190
Epoch[027] Batch [0049]/[3760]	Speed: 48.053145 samples/sec	 accuracy=53.375000	 loss=1.943580	 lr=0.010000
Epoch[027] Batch [0099]/[3760]	Speed: 67.541814 samples/sec	 accuracy=53.906250	 loss=1.949762	 lr=0.010000
Epoch[027] Batch [0149]/[3760]	Speed: 69.039405 samples/sec	 accuracy=53.739583	 loss=1.947693	 lr=0.010000
Epoch[027] Batch [0199]/[3760]	Speed: 70.343388 samples/sec	 accuracy=53.437500	 loss=1.956937	 lr=0.010000
Epoch[027] Batch [0249]/[3760]	Speed: 68.542847 samples/sec	 accuracy=53.343750	 loss=1.964893	 lr=0.010000
Epoch[027] Batch [0299]/[3760]	Speed: 70.226530 samples/sec	 accuracy=53.656250	 loss=1.952776	 lr=0.010000
Epoch[027] Batch [0349]/[3760]	Speed: 71.007105 samples/sec	 accuracy=53.589286	 loss=1.950765	 lr=0.010000
Epoch[027] Batch [0399]/[3760]	Speed: 68.743167 samples/sec	 accuracy=53.394531	 loss=1.954777	 lr=0.010000
Epoch[027] Batch [0449]/[3760]	Speed: 68.511720 samples/sec	 accuracy=53.430556	 loss=1.951918	 lr=0.010000
Epoch[027] Batch [0499]/[3760]	Speed: 70.958877 samples/sec	 accuracy=53.406250	 loss=1.954929	 lr=0.010000
Epoch[027] Batch [0549]/[3760]	Speed: 67.875330 samples/sec	 accuracy=53.400568	 loss=1.954776	 lr=0.010000
Epoch[027] Batch [0599]/[3760]	Speed: 67.926200 samples/sec	 accuracy=53.531250	 loss=1.950363	 lr=0.010000
Epoch[027] Batch [0649]/[3760]	Speed: 71.148845 samples/sec	 accuracy=53.598558	 loss=1.948093	 lr=0.010000
Epoch[027] Batch [0699]/[3760]	Speed: 69.165936 samples/sec	 accuracy=53.533482	 loss=1.951445	 lr=0.010000
Epoch[027] Batch [0749]/[3760]	Speed: 69.576286 samples/sec	 accuracy=53.545833	 loss=1.949104	 lr=0.010000
Epoch[027] Batch [0799]/[3760]	Speed: 70.743855 samples/sec	 accuracy=53.501953	 loss=1.950121	 lr=0.010000
Epoch[027] Batch [0849]/[3760]	Speed: 66.087781 samples/sec	 accuracy=53.487132	 loss=1.954198	 lr=0.010000
Epoch[027] Batch [0899]/[3760]	Speed: 71.630764 samples/sec	 accuracy=53.541667	 loss=1.953321	 lr=0.010000
Epoch[027] Batch [0949]/[3760]	Speed: 69.682365 samples/sec	 accuracy=53.611842	 loss=1.952443	 lr=0.010000
Epoch[027] Batch [0999]/[3760]	Speed: 68.155435 samples/sec	 accuracy=53.575000	 loss=1.954699	 lr=0.010000
Epoch[027] Batch [1049]/[3760]	Speed: 70.351166 samples/sec	 accuracy=53.520833	 loss=1.957603	 lr=0.010000
Epoch[027] Batch [1099]/[3760]	Speed: 71.663819 samples/sec	 accuracy=53.488636	 loss=1.958461	 lr=0.010000
Epoch[027] Batch [1149]/[3760]	Speed: 69.254048 samples/sec	 accuracy=53.498641	 loss=1.958444	 lr=0.010000
Epoch[027] Batch [1199]/[3760]	Speed: 68.830845 samples/sec	 accuracy=53.472656	 loss=1.959866	 lr=0.010000
Epoch[027] Batch [1249]/[3760]	Speed: 69.781505 samples/sec	 accuracy=53.476250	 loss=1.959828	 lr=0.010000
Epoch[027] Batch [1299]/[3760]	Speed: 70.419147 samples/sec	 accuracy=53.456731	 loss=1.961826	 lr=0.010000
Epoch[027] Batch [1349]/[3760]	Speed: 70.408273 samples/sec	 accuracy=53.475694	 loss=1.960155	 lr=0.010000
Epoch[027] Batch [1399]/[3760]	Speed: 69.522756 samples/sec	 accuracy=53.447545	 loss=1.961219	 lr=0.010000
Epoch[027] Batch [1449]/[3760]	Speed: 66.136050 samples/sec	 accuracy=53.452586	 loss=1.960994	 lr=0.010000
Epoch[027] Batch [1499]/[3760]	Speed: 71.338900 samples/sec	 accuracy=53.404167	 loss=1.962375	 lr=0.010000
Epoch[027] Batch [1549]/[3760]	Speed: 68.465013 samples/sec	 accuracy=53.421371	 loss=1.961623	 lr=0.010000
Epoch[027] Batch [1599]/[3760]	Speed: 71.122210 samples/sec	 accuracy=53.413086	 loss=1.962952	 lr=0.010000
Epoch[027] Batch [1649]/[3760]	Speed: 69.170945 samples/sec	 accuracy=53.375947	 loss=1.962655	 lr=0.010000
Epoch[027] Batch [1699]/[3760]	Speed: 70.390600 samples/sec	 accuracy=53.354779	 loss=1.963144	 lr=0.010000
Epoch[027] Batch [1749]/[3760]	Speed: 68.680506 samples/sec	 accuracy=53.383036	 loss=1.963879	 lr=0.010000
Epoch[027] Batch [1799]/[3760]	Speed: 70.234248 samples/sec	 accuracy=53.370660	 loss=1.963971	 lr=0.010000
Epoch[027] Batch [1849]/[3760]	Speed: 67.879687 samples/sec	 accuracy=53.357264	 loss=1.965197	 lr=0.010000
Epoch[027] Batch [1899]/[3760]	Speed: 70.145708 samples/sec	 accuracy=53.339638	 loss=1.966326	 lr=0.010000
Epoch[027] Batch [1949]/[3760]	Speed: 69.782750 samples/sec	 accuracy=53.326923	 loss=1.967120	 lr=0.010000
Epoch[027] Batch [1999]/[3760]	Speed: 70.679341 samples/sec	 accuracy=53.301563	 loss=1.968004	 lr=0.010000
Epoch[027] Batch [2049]/[3760]	Speed: 65.767983 samples/sec	 accuracy=53.262195	 loss=1.969132	 lr=0.010000
Epoch[027] Batch [2099]/[3760]	Speed: 72.074548 samples/sec	 accuracy=53.262649	 loss=1.969288	 lr=0.010000
Epoch[027] Batch [2149]/[3760]	Speed: 70.356639 samples/sec	 accuracy=53.265262	 loss=1.969599	 lr=0.010000
Epoch[027] Batch [2199]/[3760]	Speed: 69.648789 samples/sec	 accuracy=53.272727	 loss=1.969293	 lr=0.010000
Epoch[027] Batch [2249]/[3760]	Speed: 69.895311 samples/sec	 accuracy=53.254167	 loss=1.969225	 lr=0.010000
Epoch[027] Batch [2299]/[3760]	Speed: 70.097785 samples/sec	 accuracy=53.244565	 loss=1.970214	 lr=0.010000
Epoch[027] Batch [2349]/[3760]	Speed: 69.104099 samples/sec	 accuracy=53.228059	 loss=1.971183	 lr=0.010000
Epoch[027] Batch [2399]/[3760]	Speed: 69.037997 samples/sec	 accuracy=53.194661	 loss=1.972461	 lr=0.010000
Epoch[027] Batch [2449]/[3760]	Speed: 70.933726 samples/sec	 accuracy=53.186224	 loss=1.973736	 lr=0.010000
Epoch[027] Batch [2499]/[3760]	Speed: 69.532379 samples/sec	 accuracy=53.155625	 loss=1.975276	 lr=0.010000
Epoch[027] Batch [2549]/[3760]	Speed: 68.207326 samples/sec	 accuracy=53.153799	 loss=1.976183	 lr=0.010000
Epoch[027] Batch [2599]/[3760]	Speed: 70.365243 samples/sec	 accuracy=53.149639	 loss=1.976484	 lr=0.010000
Epoch[027] Batch [2649]/[3760]	Speed: 68.622004 samples/sec	 accuracy=53.136203	 loss=1.976971	 lr=0.010000
Epoch[027] Batch [2699]/[3760]	Speed: 70.659226 samples/sec	 accuracy=53.142361	 loss=1.977368	 lr=0.010000
Epoch[027] Batch [2749]/[3760]	Speed: 68.764425 samples/sec	 accuracy=53.114773	 loss=1.978197	 lr=0.010000
Epoch[027] Batch [2799]/[3760]	Speed: 70.879949 samples/sec	 accuracy=53.100446	 loss=1.979017	 lr=0.010000
Epoch[027] Batch [2849]/[3760]	Speed: 68.720718 samples/sec	 accuracy=53.095395	 loss=1.979595	 lr=0.010000
Epoch[027] Batch [2899]/[3760]	Speed: 68.322800 samples/sec	 accuracy=53.086207	 loss=1.979706	 lr=0.010000
Epoch[027] Batch [2949]/[3760]	Speed: 70.354168 samples/sec	 accuracy=53.075742	 loss=1.980389	 lr=0.010000
Epoch[027] Batch [2999]/[3760]	Speed: 69.643713 samples/sec	 accuracy=53.091667	 loss=1.980569	 lr=0.010000
Epoch[027] Batch [3049]/[3760]	Speed: 68.979146 samples/sec	 accuracy=53.102459	 loss=1.980276	 lr=0.010000
Epoch[027] Batch [3099]/[3760]	Speed: 70.865618 samples/sec	 accuracy=53.080141	 loss=1.980717	 lr=0.010000
Epoch[027] Batch [3149]/[3760]	Speed: 69.579558 samples/sec	 accuracy=53.078373	 loss=1.980780	 lr=0.010000
Epoch[027] Batch [3199]/[3760]	Speed: 68.683106 samples/sec	 accuracy=53.060059	 loss=1.981813	 lr=0.010000
Epoch[027] Batch [3249]/[3760]	Speed: 69.242354 samples/sec	 accuracy=53.069712	 loss=1.981414	 lr=0.010000
Epoch[027] Batch [3299]/[3760]	Speed: 69.730672 samples/sec	 accuracy=53.069602	 loss=1.981691	 lr=0.010000
Epoch[027] Batch [3349]/[3760]	Speed: 69.902079 samples/sec	 accuracy=53.071828	 loss=1.982235	 lr=0.010000
Epoch[027] Batch [3399]/[3760]	Speed: 69.767778 samples/sec	 accuracy=53.083640	 loss=1.982218	 lr=0.010000
Epoch[027] Batch [3449]/[3760]	Speed: 70.461672 samples/sec	 accuracy=53.062953	 loss=1.983516	 lr=0.010000
Epoch[027] Batch [3499]/[3760]	Speed: 70.223655 samples/sec	 accuracy=53.054464	 loss=1.983796	 lr=0.010000
Epoch[027] Batch [3549]/[3760]	Speed: 69.418225 samples/sec	 accuracy=53.073944	 loss=1.982870	 lr=0.010000
Epoch[027] Batch [3599]/[3760]	Speed: 68.378724 samples/sec	 accuracy=53.075955	 loss=1.982915	 lr=0.010000
Epoch[027] Batch [3649]/[3760]	Speed: 69.691578 samples/sec	 accuracy=53.079623	 loss=1.982500	 lr=0.010000
Epoch[027] Batch [3699]/[3760]	Speed: 71.452143 samples/sec	 accuracy=53.059966	 loss=1.983419	 lr=0.010000
Epoch[027] Batch [3749]/[3760]	Speed: 76.648517 samples/sec	 accuracy=53.063750	 loss=1.983664	 lr=0.010000
Batch [0049]/[0303]: acc-top1=52.687500 acc-top5=77.875000
Batch [0099]/[0303]: acc-top1=52.718750 acc-top5=77.890625
Batch [0149]/[0303]: acc-top1=53.218750 acc-top5=78.322917
Batch [0199]/[0303]: acc-top1=53.796875 acc-top5=78.515625
Batch [0249]/[0303]: acc-top1=53.868750 acc-top5=78.656250
Batch [0299]/[0303]: acc-top1=53.677083 acc-top5=78.484375
[Epoch 027] training: accuracy=53.063497	 loss=1.983529
[Epoch 027] speed: 69 samples/sec	time cost: 3767.793928
[Epoch 027] validation: acc-top1=53.681931 acc-top5=78.491130 loss=2.056067
Epoch[028] Batch [0049]/[3760]	Speed: 47.403273 samples/sec	 accuracy=54.625000	 loss=1.924072	 lr=0.010000
Epoch[028] Batch [0099]/[3760]	Speed: 70.114746 samples/sec	 accuracy=54.125000	 loss=1.948630	 lr=0.010000
Epoch[028] Batch [0149]/[3760]	Speed: 69.028644 samples/sec	 accuracy=54.083333	 loss=1.945456	 lr=0.010000
Epoch[028] Batch [0199]/[3760]	Speed: 69.064304 samples/sec	 accuracy=53.960937	 loss=1.955037	 lr=0.010000
Epoch[028] Batch [0249]/[3760]	Speed: 69.777899 samples/sec	 accuracy=54.175000	 loss=1.948711	 lr=0.010000
Epoch[028] Batch [0299]/[3760]	Speed: 70.052383 samples/sec	 accuracy=54.088542	 loss=1.945584	 lr=0.010000
Epoch[028] Batch [0349]/[3760]	Speed: 70.630673 samples/sec	 accuracy=54.183036	 loss=1.943704	 lr=0.010000
Epoch[028] Batch [0399]/[3760]	Speed: 69.995696 samples/sec	 accuracy=54.277344	 loss=1.935424	 lr=0.010000
Epoch[028] Batch [0449]/[3760]	Speed: 69.000400 samples/sec	 accuracy=54.371528	 loss=1.930554	 lr=0.010000
Epoch[028] Batch [0499]/[3760]	Speed: 70.658745 samples/sec	 accuracy=54.265625	 loss=1.934020	 lr=0.010000
Epoch[028] Batch [0549]/[3760]	Speed: 67.613249 samples/sec	 accuracy=54.122159	 loss=1.938909	 lr=0.010000
Epoch[028] Batch [0599]/[3760]	Speed: 69.826634 samples/sec	 accuracy=54.052083	 loss=1.942303	 lr=0.010000
Epoch[028] Batch [0649]/[3760]	Speed: 70.004063 samples/sec	 accuracy=54.127404	 loss=1.942190	 lr=0.010000
Epoch[028] Batch [0699]/[3760]	Speed: 70.225523 samples/sec	 accuracy=54.087054	 loss=1.941962	 lr=0.010000
Epoch[028] Batch [0749]/[3760]	Speed: 69.831861 samples/sec	 accuracy=54.087500	 loss=1.939979	 lr=0.010000
Epoch[028] Batch [0799]/[3760]	Speed: 69.367106 samples/sec	 accuracy=54.068359	 loss=1.942803	 lr=0.010000
Epoch[028] Batch [0849]/[3760]	Speed: 70.014611 samples/sec	 accuracy=54.068015	 loss=1.945033	 lr=0.010000
Epoch[028] Batch [0899]/[3760]	Speed: 69.778124 samples/sec	 accuracy=54.059028	 loss=1.944505	 lr=0.010000
Epoch[028] Batch [0949]/[3760]	Speed: 69.086629 samples/sec	 accuracy=53.985197	 loss=1.947383	 lr=0.010000
Epoch[028] Batch [0999]/[3760]	Speed: 70.085235 samples/sec	 accuracy=53.992188	 loss=1.948269	 lr=0.010000
Epoch[028] Batch [1049]/[3760]	Speed: 69.597600 samples/sec	 accuracy=53.946429	 loss=1.949532	 lr=0.010000
Epoch[028] Batch [1099]/[3760]	Speed: 68.887704 samples/sec	 accuracy=53.981534	 loss=1.947458	 lr=0.010000
Epoch[028] Batch [1149]/[3760]	Speed: 70.246931 samples/sec	 accuracy=53.941576	 loss=1.950118	 lr=0.010000
Epoch[028] Batch [1199]/[3760]	Speed: 69.990069 samples/sec	 accuracy=53.860677	 loss=1.954512	 lr=0.010000
Epoch[028] Batch [1249]/[3760]	Speed: 69.698797 samples/sec	 accuracy=53.797500	 loss=1.956195	 lr=0.010000
Epoch[028] Batch [1299]/[3760]	Speed: 70.083988 samples/sec	 accuracy=53.753606	 loss=1.958712	 lr=0.010000
Epoch[028] Batch [1349]/[3760]	Speed: 69.955102 samples/sec	 accuracy=53.712963	 loss=1.961416	 lr=0.010000
Epoch[028] Batch [1399]/[3760]	Speed: 69.492433 samples/sec	 accuracy=53.693080	 loss=1.963299	 lr=0.010000
Epoch[028] Batch [1449]/[3760]	Speed: 70.838315 samples/sec	 accuracy=53.657328	 loss=1.964137	 lr=0.010000
Epoch[028] Batch [1499]/[3760]	Speed: 67.625650 samples/sec	 accuracy=53.627083	 loss=1.967259	 lr=0.010000
Epoch[028] Batch [1549]/[3760]	Speed: 71.273406 samples/sec	 accuracy=53.658266	 loss=1.965055	 lr=0.010000
Epoch[028] Batch [1599]/[3760]	Speed: 68.697554 samples/sec	 accuracy=53.632813	 loss=1.965593	 lr=0.010000
Epoch[028] Batch [1649]/[3760]	Speed: 70.543894 samples/sec	 accuracy=53.568182	 loss=1.968040	 lr=0.010000
Epoch[028] Batch [1699]/[3760]	Speed: 69.634263 samples/sec	 accuracy=53.571691	 loss=1.967389	 lr=0.010000
Epoch[028] Batch [1749]/[3760]	Speed: 69.700998 samples/sec	 accuracy=53.537500	 loss=1.967821	 lr=0.010000
Epoch[028] Batch [1799]/[3760]	Speed: 69.307618 samples/sec	 accuracy=53.561632	 loss=1.967275	 lr=0.010000
Epoch[028] Batch [1849]/[3760]	Speed: 68.305944 samples/sec	 accuracy=53.549831	 loss=1.967468	 lr=0.010000
Epoch[028] Batch [1899]/[3760]	Speed: 69.868958 samples/sec	 accuracy=53.577303	 loss=1.965904	 lr=0.010000
Epoch[028] Batch [1949]/[3760]	Speed: 70.033166 samples/sec	 accuracy=53.573718	 loss=1.965329	 lr=0.010000
Epoch[028] Batch [1999]/[3760]	Speed: 70.722348 samples/sec	 accuracy=53.569531	 loss=1.964546	 lr=0.010000
Epoch[028] Batch [2049]/[3760]	Speed: 69.969640 samples/sec	 accuracy=53.538872	 loss=1.965754	 lr=0.010000
Epoch[028] Batch [2099]/[3760]	Speed: 68.663569 samples/sec	 accuracy=53.539435	 loss=1.966283	 lr=0.010000
Epoch[028] Batch [2149]/[3760]	Speed: 69.158597 samples/sec	 accuracy=53.539244	 loss=1.966427	 lr=0.010000
Epoch[028] Batch [2199]/[3760]	Speed: 70.072066 samples/sec	 accuracy=53.523438	 loss=1.967569	 lr=0.010000
Epoch[028] Batch [2249]/[3760]	Speed: 70.062100 samples/sec	 accuracy=53.500000	 loss=1.969062	 lr=0.010000
Epoch[028] Batch [2299]/[3760]	Speed: 70.311684 samples/sec	 accuracy=53.502717	 loss=1.968683	 lr=0.010000
Epoch[028] Batch [2349]/[3760]	Speed: 69.344536 samples/sec	 accuracy=53.501995	 loss=1.968643	 lr=0.010000
Epoch[028] Batch [2399]/[3760]	Speed: 69.710230 samples/sec	 accuracy=53.508464	 loss=1.968744	 lr=0.010000
Epoch[028] Batch [2449]/[3760]	Speed: 68.385228 samples/sec	 accuracy=53.500638	 loss=1.969241	 lr=0.010000
Epoch[028] Batch [2499]/[3760]	Speed: 71.011194 samples/sec	 accuracy=53.465000	 loss=1.970450	 lr=0.010000
Epoch[028] Batch [2549]/[3760]	Speed: 69.024572 samples/sec	 accuracy=53.473039	 loss=1.970195	 lr=0.010000
Epoch[028] Batch [2599]/[3760]	Speed: 68.934836 samples/sec	 accuracy=53.439303	 loss=1.972279	 lr=0.010000
Epoch[028] Batch [2649]/[3760]	Speed: 69.321230 samples/sec	 accuracy=53.439858	 loss=1.973141	 lr=0.010000
Epoch[028] Batch [2699]/[3760]	Speed: 70.096658 samples/sec	 accuracy=53.406829	 loss=1.974625	 lr=0.010000
Epoch[028] Batch [2749]/[3760]	Speed: 69.179455 samples/sec	 accuracy=53.388068	 loss=1.976010	 lr=0.010000
Epoch[028] Batch [2799]/[3760]	Speed: 71.280820 samples/sec	 accuracy=53.318080	 loss=1.978998	 lr=0.010000
Epoch[028] Batch [2849]/[3760]	Speed: 68.564343 samples/sec	 accuracy=53.304276	 loss=1.980131	 lr=0.010000
Epoch[028] Batch [2899]/[3760]	Speed: 71.573678 samples/sec	 accuracy=53.274784	 loss=1.981532	 lr=0.010000
Epoch[028] Batch [2949]/[3760]	Speed: 68.783114 samples/sec	 accuracy=53.313559	 loss=1.979621	 lr=0.010000
Epoch[028] Batch [2999]/[3760]	Speed: 68.990735 samples/sec	 accuracy=53.317187	 loss=1.979479	 lr=0.010000
Epoch[028] Batch [3049]/[3760]	Speed: 71.132588 samples/sec	 accuracy=53.324283	 loss=1.979282	 lr=0.010000
Epoch[028] Batch [3099]/[3760]	Speed: 70.321954 samples/sec	 accuracy=53.332157	 loss=1.978801	 lr=0.010000
Epoch[028] Batch [3149]/[3760]	Speed: 68.832615 samples/sec	 accuracy=53.318452	 loss=1.980110	 lr=0.010000
Epoch[028] Batch [3199]/[3760]	Speed: 68.565251 samples/sec	 accuracy=53.312012	 loss=1.980764	 lr=0.010000
Epoch[028] Batch [3249]/[3760]	Speed: 71.352431 samples/sec	 accuracy=53.298077	 loss=1.981040	 lr=0.010000
Epoch[028] Batch [3299]/[3760]	Speed: 70.424813 samples/sec	 accuracy=53.272727	 loss=1.981923	 lr=0.010000
Epoch[028] Batch [3349]/[3760]	Speed: 69.123223 samples/sec	 accuracy=53.277985	 loss=1.981782	 lr=0.010000
Epoch[028] Batch [3399]/[3760]	Speed: 69.247581 samples/sec	 accuracy=53.250919	 loss=1.982868	 lr=0.010000
Epoch[028] Batch [3449]/[3760]	Speed: 70.255907 samples/sec	 accuracy=53.230978	 loss=1.983596	 lr=0.010000
Epoch[028] Batch [3499]/[3760]	Speed: 69.787030 samples/sec	 accuracy=53.249107	 loss=1.983516	 lr=0.010000
Epoch[028] Batch [3549]/[3760]	Speed: 69.051637 samples/sec	 accuracy=53.251761	 loss=1.983004	 lr=0.010000
Epoch[028] Batch [3599]/[3760]	Speed: 70.827257 samples/sec	 accuracy=53.226128	 loss=1.984374	 lr=0.010000
Epoch[028] Batch [3649]/[3760]	Speed: 69.383286 samples/sec	 accuracy=53.212329	 loss=1.984887	 lr=0.010000
Epoch[028] Batch [3699]/[3760]	Speed: 69.600310 samples/sec	 accuracy=53.226774	 loss=1.984384	 lr=0.010000
Epoch[028] Batch [3749]/[3760]	Speed: 76.493367 samples/sec	 accuracy=53.237083	 loss=1.983406	 lr=0.010000
Batch [0049]/[0303]: acc-top1=52.687500 acc-top5=78.000000
Batch [0099]/[0303]: acc-top1=52.421875 acc-top5=77.640625
Batch [0149]/[0303]: acc-top1=53.135417 acc-top5=78.072917
Batch [0199]/[0303]: acc-top1=53.234375 acc-top5=78.367188
Batch [0249]/[0303]: acc-top1=53.431250 acc-top5=78.525000
Batch [0299]/[0303]: acc-top1=53.302083 acc-top5=78.307292
[Epoch 028] training: accuracy=53.234707	 loss=1.983058
[Epoch 028] speed: 69 samples/sec	time cost: 3760.504089
[Epoch 028] validation: acc-top1=53.290017 acc-top5=78.295173 loss=2.063219
Epoch[029] Batch [0049]/[3759]	Speed: 48.803916 samples/sec	 accuracy=55.125000	 loss=1.918762	 lr=0.010000
Epoch[029] Batch [0099]/[3759]	Speed: 69.116499 samples/sec	 accuracy=54.562500	 loss=1.930891	 lr=0.010000
Epoch[029] Batch [0149]/[3759]	Speed: 67.965809 samples/sec	 accuracy=54.458333	 loss=1.929653	 lr=0.010000
Epoch[029] Batch [0199]/[3759]	Speed: 69.208543 samples/sec	 accuracy=54.781250	 loss=1.922455	 lr=0.010000
Epoch[029] Batch [0249]/[3759]	Speed: 69.740054 samples/sec	 accuracy=54.506250	 loss=1.931044	 lr=0.010000
Epoch[029] Batch [0299]/[3759]	Speed: 69.842165 samples/sec	 accuracy=54.369792	 loss=1.927693	 lr=0.010000
Epoch[029] Batch [0349]/[3759]	Speed: 69.442073 samples/sec	 accuracy=54.098214	 loss=1.933191	 lr=0.010000
Epoch[029] Batch [0399]/[3759]	Speed: 68.544247 samples/sec	 accuracy=54.128906	 loss=1.930876	 lr=0.010000
Epoch[029] Batch [0449]/[3759]	Speed: 68.813519 samples/sec	 accuracy=53.986111	 loss=1.936842	 lr=0.010000
Epoch[029] Batch [0499]/[3759]	Speed: 72.322447 samples/sec	 accuracy=53.887500	 loss=1.946110	 lr=0.010000
Epoch[029] Batch [0549]/[3759]	Speed: 69.857497 samples/sec	 accuracy=53.855114	 loss=1.951428	 lr=0.010000
Epoch[029] Batch [0599]/[3759]	Speed: 69.031454 samples/sec	 accuracy=53.955729	 loss=1.945764	 lr=0.010000
Epoch[029] Batch [0649]/[3759]	Speed: 71.177359 samples/sec	 accuracy=53.973558	 loss=1.947560	 lr=0.010000
Epoch[029] Batch [0699]/[3759]	Speed: 69.133997 samples/sec	 accuracy=53.921875	 loss=1.949736	 lr=0.010000
Epoch[029] Batch [0749]/[3759]	Speed: 69.629005 samples/sec	 accuracy=53.958333	 loss=1.948541	 lr=0.010000
Epoch[029] Batch [0799]/[3759]	Speed: 69.837486 samples/sec	 accuracy=53.927734	 loss=1.951955	 lr=0.010000
Epoch[029] Batch [0849]/[3759]	Speed: 68.494928 samples/sec	 accuracy=53.974265	 loss=1.948713	 lr=0.010000
Epoch[029] Batch [0899]/[3759]	Speed: 70.385731 samples/sec	 accuracy=53.980903	 loss=1.947591	 lr=0.010000
Epoch[029] Batch [0949]/[3759]	Speed: 69.856237 samples/sec	 accuracy=53.963816	 loss=1.948255	 lr=0.010000
Epoch[029] Batch [0999]/[3759]	Speed: 69.605884 samples/sec	 accuracy=53.926562	 loss=1.949956	 lr=0.010000
Epoch[029] Batch [1049]/[3759]	Speed: 68.296196 samples/sec	 accuracy=53.943452	 loss=1.949377	 lr=0.010000
Epoch[029] Batch [1099]/[3759]	Speed: 70.786592 samples/sec	 accuracy=53.903409	 loss=1.949730	 lr=0.010000
Epoch[029] Batch [1149]/[3759]	Speed: 69.020016 samples/sec	 accuracy=53.904891	 loss=1.949722	 lr=0.010000
Epoch[029] Batch [1199]/[3759]	Speed: 69.614100 samples/sec	 accuracy=53.824219	 loss=1.952816	 lr=0.010000
Epoch[029] Batch [1249]/[3759]	Speed: 69.545977 samples/sec	 accuracy=53.826250	 loss=1.953716	 lr=0.010000
Epoch[029] Batch [1299]/[3759]	Speed: 70.870923 samples/sec	 accuracy=53.820913	 loss=1.953732	 lr=0.010000
Epoch[029] Batch [1349]/[3759]	Speed: 69.330661 samples/sec	 accuracy=53.763889	 loss=1.955984	 lr=0.010000
Epoch[029] Batch [1399]/[3759]	Speed: 69.149332 samples/sec	 accuracy=53.691964	 loss=1.957393	 lr=0.010000
Epoch[029] Batch [1449]/[3759]	Speed: 70.619798 samples/sec	 accuracy=53.703664	 loss=1.957743	 lr=0.010000
Epoch[029] Batch [1499]/[3759]	Speed: 67.580765 samples/sec	 accuracy=53.730208	 loss=1.957344	 lr=0.010000
Epoch[029] Batch [1549]/[3759]	Speed: 70.223414 samples/sec	 accuracy=53.697581	 loss=1.958298	 lr=0.010000
Epoch[029] Batch [1599]/[3759]	Speed: 71.649019 samples/sec	 accuracy=53.713867	 loss=1.958005	 lr=0.010000
Epoch[029] Batch [1649]/[3759]	Speed: 68.451049 samples/sec	 accuracy=53.728220	 loss=1.958030	 lr=0.010000
Epoch[029] Batch [1699]/[3759]	Speed: 70.785915 samples/sec	 accuracy=53.737132	 loss=1.959092	 lr=0.010000
Epoch[029] Batch [1749]/[3759]	Speed: 69.622812 samples/sec	 accuracy=53.749107	 loss=1.959181	 lr=0.010000
Epoch[029] Batch [1799]/[3759]	Speed: 68.426688 samples/sec	 accuracy=53.759549	 loss=1.959121	 lr=0.010000
Epoch[029] Batch [1849]/[3759]	Speed: 70.490490 samples/sec	 accuracy=53.750845	 loss=1.959526	 lr=0.010000
Epoch[029] Batch [1899]/[3759]	Speed: 70.406713 samples/sec	 accuracy=53.758224	 loss=1.958972	 lr=0.010000
Epoch[029] Batch [1949]/[3759]	Speed: 68.436626 samples/sec	 accuracy=53.752404	 loss=1.958754	 lr=0.010000
Epoch[029] Batch [1999]/[3759]	Speed: 70.964454 samples/sec	 accuracy=53.763281	 loss=1.958517	 lr=0.010000
Epoch[029] Batch [2049]/[3759]	Speed: 68.210977 samples/sec	 accuracy=53.699695	 loss=1.960535	 lr=0.010000
Epoch[029] Batch [2099]/[3759]	Speed: 70.856669 samples/sec	 accuracy=53.651042	 loss=1.962549	 lr=0.010000
Epoch[029] Batch [2149]/[3759]	Speed: 68.973629 samples/sec	 accuracy=53.678779	 loss=1.961651	 lr=0.010000
Epoch[029] Batch [2199]/[3759]	Speed: 71.016818 samples/sec	 accuracy=53.686080	 loss=1.960807	 lr=0.010000
Epoch[029] Batch [2249]/[3759]	Speed: 70.128247 samples/sec	 accuracy=53.700000	 loss=1.959643	 lr=0.010000
Epoch[029] Batch [2299]/[3759]	Speed: 70.166254 samples/sec	 accuracy=53.677989	 loss=1.959828	 lr=0.010000
Epoch[029] Batch [2349]/[3759]	Speed: 69.442070 samples/sec	 accuracy=53.680851	 loss=1.960103	 lr=0.010000
Epoch[029] Batch [2399]/[3759]	Speed: 67.803405 samples/sec	 accuracy=53.688151	 loss=1.959592	 lr=0.010000
Epoch[029] Batch [2449]/[3759]	Speed: 70.412810 samples/sec	 accuracy=53.655612	 loss=1.961320	 lr=0.010000
Epoch[029] Batch [2499]/[3759]	Speed: 69.398003 samples/sec	 accuracy=53.659375	 loss=1.960553	 lr=0.010000
Epoch[029] Batch [2549]/[3759]	Speed: 71.039193 samples/sec	 accuracy=53.643382	 loss=1.961259	 lr=0.010000
Epoch[029] Batch [2599]/[3759]	Speed: 69.120677 samples/sec	 accuracy=53.660457	 loss=1.960334	 lr=0.010000
Epoch[029] Batch [2649]/[3759]	Speed: 69.947518 samples/sec	 accuracy=53.658608	 loss=1.960691	 lr=0.010000
Epoch[029] Batch [2699]/[3759]	Speed: 69.535036 samples/sec	 accuracy=53.630787	 loss=1.962043	 lr=0.010000
Epoch[029] Batch [2749]/[3759]	Speed: 68.738010 samples/sec	 accuracy=53.646023	 loss=1.961419	 lr=0.010000
Epoch[029] Batch [2799]/[3759]	Speed: 69.777761 samples/sec	 accuracy=53.659040	 loss=1.961134	 lr=0.010000
Epoch[029] Batch [2849]/[3759]	Speed: 70.488409 samples/sec	 accuracy=53.671053	 loss=1.961017	 lr=0.010000
Epoch[029] Batch [2899]/[3759]	Speed: 69.721871 samples/sec	 accuracy=53.644397	 loss=1.962247	 lr=0.010000
Epoch[029] Batch [2949]/[3759]	Speed: 69.148873 samples/sec	 accuracy=53.638242	 loss=1.962792	 lr=0.010000
Epoch[029] Batch [2999]/[3759]	Speed: 69.194027 samples/sec	 accuracy=53.631250	 loss=1.963135	 lr=0.010000
Epoch[029] Batch [3049]/[3759]	Speed: 70.190573 samples/sec	 accuracy=53.625000	 loss=1.962875	 lr=0.010000
Epoch[029] Batch [3099]/[3759]	Speed: 69.641279 samples/sec	 accuracy=53.605847	 loss=1.963276	 lr=0.010000
Epoch[029] Batch [3149]/[3759]	Speed: 68.819989 samples/sec	 accuracy=53.612599	 loss=1.963600	 lr=0.010000
Epoch[029] Batch [3199]/[3759]	Speed: 69.783524 samples/sec	 accuracy=53.601562	 loss=1.963943	 lr=0.010000
Epoch[029] Batch [3249]/[3759]	Speed: 68.088725 samples/sec	 accuracy=53.593269	 loss=1.964317	 lr=0.010000
Epoch[029] Batch [3299]/[3759]	Speed: 69.720044 samples/sec	 accuracy=53.583807	 loss=1.965080	 lr=0.010000
Epoch[029] Batch [3349]/[3759]	Speed: 68.087631 samples/sec	 accuracy=53.600280	 loss=1.964980	 lr=0.010000
Epoch[029] Batch [3399]/[3759]	Speed: 70.732605 samples/sec	 accuracy=53.578585	 loss=1.965428	 lr=0.010000
Epoch[029] Batch [3449]/[3759]	Speed: 70.579531 samples/sec	 accuracy=53.560236	 loss=1.966508	 lr=0.010000
Epoch[029] Batch [3499]/[3759]	Speed: 68.851419 samples/sec	 accuracy=53.537054	 loss=1.967175	 lr=0.010000
Epoch[029] Batch [3549]/[3759]	Speed: 71.213219 samples/sec	 accuracy=53.538732	 loss=1.967410	 lr=0.010000
Epoch[029] Batch [3599]/[3759]	Speed: 64.548723 samples/sec	 accuracy=53.532118	 loss=1.967524	 lr=0.010000
Epoch[029] Batch [3649]/[3759]	Speed: 74.303351 samples/sec	 accuracy=53.534247	 loss=1.967670	 lr=0.010000
Epoch[029] Batch [3699]/[3759]	Speed: 67.747382 samples/sec	 accuracy=53.535473	 loss=1.967853	 lr=0.010000
Epoch[029] Batch [3749]/[3759]	Speed: 77.997232 samples/sec	 accuracy=53.536250	 loss=1.967550	 lr=0.010000
Batch [0049]/[0303]: acc-top1=52.687500 acc-top5=77.531250
Batch [0099]/[0303]: acc-top1=53.125000 acc-top5=78.093750
Batch [0149]/[0303]: acc-top1=53.979167 acc-top5=78.645833
Batch [0199]/[0303]: acc-top1=54.234375 acc-top5=78.664062
Batch [0249]/[0303]: acc-top1=54.318750 acc-top5=78.843750
Batch [0299]/[0303]: acc-top1=54.140625 acc-top5=78.640625
[Epoch 029] training: accuracy=53.538591	 loss=1.967556
[Epoch 029] speed: 69 samples/sec	time cost: 3761.649311
[Epoch 029] validation: acc-top1=54.099629 acc-top5=78.604579 loss=2.030927
Epoch[030] Batch [0049]/[3760]	Speed: 48.013370 samples/sec	 accuracy=53.750000	 loss=1.927752	 lr=0.010000
Epoch[030] Batch [0099]/[3760]	Speed: 69.191895 samples/sec	 accuracy=53.890625	 loss=1.950653	 lr=0.010000
Epoch[030] Batch [0149]/[3760]	Speed: 68.732184 samples/sec	 accuracy=53.989583	 loss=1.948207	 lr=0.010000
Epoch[030] Batch [0199]/[3760]	Speed: 66.754426 samples/sec	 accuracy=54.085938	 loss=1.945736	 lr=0.010000
Epoch[030] Batch [0249]/[3760]	Speed: 70.099745 samples/sec	 accuracy=54.381250	 loss=1.934753	 lr=0.010000
Epoch[030] Batch [0299]/[3760]	Speed: 71.738650 samples/sec	 accuracy=54.531250	 loss=1.930237	 lr=0.010000
Epoch[030] Batch [0349]/[3760]	Speed: 67.346533 samples/sec	 accuracy=54.625000	 loss=1.927080	 lr=0.010000
Epoch[030] Batch [0399]/[3760]	Speed: 69.629868 samples/sec	 accuracy=54.437500	 loss=1.931298	 lr=0.010000
Epoch[030] Batch [0449]/[3760]	Speed: 71.990099 samples/sec	 accuracy=54.326389	 loss=1.936324	 lr=0.010000
Epoch[030] Batch [0499]/[3760]	Speed: 69.260007 samples/sec	 accuracy=54.237500	 loss=1.936884	 lr=0.010000
Epoch[030] Batch [0549]/[3760]	Speed: 70.515986 samples/sec	 accuracy=54.261364	 loss=1.934086	 lr=0.010000
Epoch[030] Batch [0599]/[3760]	Speed: 69.393771 samples/sec	 accuracy=54.197917	 loss=1.936346	 lr=0.010000
Epoch[030] Batch [0649]/[3760]	Speed: 67.647369 samples/sec	 accuracy=54.100962	 loss=1.938993	 lr=0.010000
Epoch[030] Batch [0699]/[3760]	Speed: 69.534305 samples/sec	 accuracy=54.087054	 loss=1.941877	 lr=0.010000
Epoch[030] Batch [0749]/[3760]	Speed: 69.364752 samples/sec	 accuracy=54.004167	 loss=1.942874	 lr=0.010000
Epoch[030] Batch [0799]/[3760]	Speed: 71.246242 samples/sec	 accuracy=53.970703	 loss=1.946143	 lr=0.010000
Epoch[030] Batch [0849]/[3760]	Speed: 69.688382 samples/sec	 accuracy=53.941176	 loss=1.948607	 lr=0.010000
Epoch[030] Batch [0899]/[3760]	Speed: 69.558680 samples/sec	 accuracy=53.878472	 loss=1.950300	 lr=0.010000
Epoch[030] Batch [0949]/[3760]	Speed: 69.399011 samples/sec	 accuracy=53.896382	 loss=1.948742	 lr=0.010000
Epoch[030] Batch [0999]/[3760]	Speed: 70.704610 samples/sec	 accuracy=53.815625	 loss=1.951514	 lr=0.010000
Epoch[030] Batch [1049]/[3760]	Speed: 67.620901 samples/sec	 accuracy=53.787202	 loss=1.953820	 lr=0.010000
Epoch[030] Batch [1099]/[3760]	Speed: 70.706312 samples/sec	 accuracy=53.755682	 loss=1.956195	 lr=0.010000
Epoch[030] Batch [1149]/[3760]	Speed: 69.523088 samples/sec	 accuracy=53.735054	 loss=1.955524	 lr=0.010000
Epoch[030] Batch [1199]/[3760]	Speed: 69.615233 samples/sec	 accuracy=53.776042	 loss=1.954004	 lr=0.010000
Epoch[030] Batch [1249]/[3760]	Speed: 69.655308 samples/sec	 accuracy=53.752500	 loss=1.956036	 lr=0.010000
Epoch[030] Batch [1299]/[3760]	Speed: 69.461924 samples/sec	 accuracy=53.762019	 loss=1.955348	 lr=0.010000
Epoch[030] Batch [1349]/[3760]	Speed: 70.113453 samples/sec	 accuracy=53.813657	 loss=1.953292	 lr=0.010000
Epoch[030] Batch [1399]/[3760]	Speed: 69.145425 samples/sec	 accuracy=53.799107	 loss=1.953002	 lr=0.010000
Epoch[030] Batch [1449]/[3760]	Speed: 68.585107 samples/sec	 accuracy=53.831897	 loss=1.952433	 lr=0.010000
Epoch[030] Batch [1499]/[3760]	Speed: 71.009613 samples/sec	 accuracy=53.797917	 loss=1.952039	 lr=0.010000
Epoch[030] Batch [1549]/[3760]	Speed: 68.895272 samples/sec	 accuracy=53.806452	 loss=1.953086	 lr=0.010000
Epoch[030] Batch [1599]/[3760]	Speed: 70.851926 samples/sec	 accuracy=53.790039	 loss=1.953423	 lr=0.010000
Epoch[030] Batch [1649]/[3760]	Speed: 67.054900 samples/sec	 accuracy=53.775568	 loss=1.954498	 lr=0.010000
Epoch[030] Batch [1699]/[3760]	Speed: 72.306075 samples/sec	 accuracy=53.746324	 loss=1.955526	 lr=0.010000
Epoch[030] Batch [1749]/[3760]	Speed: 68.063508 samples/sec	 accuracy=53.709821	 loss=1.955809	 lr=0.010000
Epoch[030] Batch [1799]/[3760]	Speed: 70.768628 samples/sec	 accuracy=53.698785	 loss=1.956142	 lr=0.010000
Epoch[030] Batch [1849]/[3760]	Speed: 67.368187 samples/sec	 accuracy=53.662162	 loss=1.957829	 lr=0.010000
Epoch[030] Batch [1899]/[3760]	Speed: 72.096878 samples/sec	 accuracy=53.628289	 loss=1.958769	 lr=0.010000
Epoch[030] Batch [1949]/[3760]	Speed: 69.003166 samples/sec	 accuracy=53.623397	 loss=1.959208	 lr=0.010000
Epoch[030] Batch [1999]/[3760]	Speed: 68.232322 samples/sec	 accuracy=53.610156	 loss=1.959906	 lr=0.010000
Epoch[030] Batch [2049]/[3760]	Speed: 71.185405 samples/sec	 accuracy=53.594512	 loss=1.960624	 lr=0.010000
Epoch[030] Batch [2099]/[3760]	Speed: 69.115788 samples/sec	 accuracy=53.616815	 loss=1.960326	 lr=0.010000
Epoch[030] Batch [2149]/[3760]	Speed: 68.709927 samples/sec	 accuracy=53.658430	 loss=1.959011	 lr=0.010000
Epoch[030] Batch [2199]/[3760]	Speed: 70.483243 samples/sec	 accuracy=53.651989	 loss=1.958732	 lr=0.010000
Epoch[030] Batch [2249]/[3760]	Speed: 69.220297 samples/sec	 accuracy=53.649306	 loss=1.959011	 lr=0.010000
Epoch[030] Batch [2299]/[3760]	Speed: 68.548173 samples/sec	 accuracy=53.639946	 loss=1.959536	 lr=0.010000
Epoch[030] Batch [2349]/[3760]	Speed: 70.771447 samples/sec	 accuracy=53.662234	 loss=1.958904	 lr=0.010000
Epoch[030] Batch [2399]/[3760]	Speed: 70.758088 samples/sec	 accuracy=53.655599	 loss=1.958502	 lr=0.010000
Epoch[030] Batch [2449]/[3760]	Speed: 68.739101 samples/sec	 accuracy=53.636480	 loss=1.959520	 lr=0.010000
Epoch[030] Batch [2499]/[3760]	Speed: 70.218651 samples/sec	 accuracy=53.626875	 loss=1.960285	 lr=0.010000
Epoch[030] Batch [2549]/[3760]	Speed: 69.221817 samples/sec	 accuracy=53.622549	 loss=1.960328	 lr=0.010000
Epoch[030] Batch [2599]/[3760]	Speed: 68.459692 samples/sec	 accuracy=53.637019	 loss=1.959774	 lr=0.010000
Epoch[030] Batch [2649]/[3760]	Speed: 70.889417 samples/sec	 accuracy=53.637972	 loss=1.960968	 lr=0.010000
Epoch[030] Batch [2699]/[3760]	Speed: 68.332235 samples/sec	 accuracy=53.613426	 loss=1.961968	 lr=0.010000
Epoch[030] Batch [2749]/[3760]	Speed: 70.789514 samples/sec	 accuracy=53.640341	 loss=1.960961	 lr=0.010000
Epoch[030] Batch [2799]/[3760]	Speed: 69.632929 samples/sec	 accuracy=53.646763	 loss=1.961037	 lr=0.010000
Epoch[030] Batch [2849]/[3760]	Speed: 68.713068 samples/sec	 accuracy=53.648575	 loss=1.961503	 lr=0.010000
Epoch[030] Batch [2899]/[3760]	Speed: 71.352329 samples/sec	 accuracy=53.652478	 loss=1.962324	 lr=0.010000
Epoch[030] Batch [2949]/[3760]	Speed: 67.608755 samples/sec	 accuracy=53.643538	 loss=1.962675	 lr=0.010000
Epoch[030] Batch [2999]/[3760]	Speed: 71.640078 samples/sec	 accuracy=53.650000	 loss=1.962011	 lr=0.010000
Epoch[030] Batch [3049]/[3760]	Speed: 69.032787 samples/sec	 accuracy=53.646516	 loss=1.962508	 lr=0.010000
Epoch[030] Batch [3099]/[3760]	Speed: 69.206106 samples/sec	 accuracy=53.643145	 loss=1.961745	 lr=0.010000
Epoch[030] Batch [3149]/[3760]	Speed: 70.999904 samples/sec	 accuracy=53.636905	 loss=1.961853	 lr=0.010000
Epoch[030] Batch [3199]/[3760]	Speed: 70.332989 samples/sec	 accuracy=53.625977	 loss=1.962234	 lr=0.010000
Epoch[030] Batch [3249]/[3760]	Speed: 68.345307 samples/sec	 accuracy=53.640865	 loss=1.962582	 lr=0.010000
Epoch[030] Batch [3299]/[3760]	Speed: 71.114794 samples/sec	 accuracy=53.605587	 loss=1.964027	 lr=0.010000
Epoch[030] Batch [3349]/[3760]	Speed: 68.690873 samples/sec	 accuracy=53.594216	 loss=1.964218	 lr=0.010000
Epoch[030] Batch [3399]/[3760]	Speed: 71.201680 samples/sec	 accuracy=53.630055	 loss=1.962842	 lr=0.010000
Epoch[030] Batch [3449]/[3760]	Speed: 68.598548 samples/sec	 accuracy=53.639493	 loss=1.962305	 lr=0.010000
Epoch[030] Batch [3499]/[3760]	Speed: 69.729209 samples/sec	 accuracy=53.647768	 loss=1.962278	 lr=0.010000
Epoch[030] Batch [3549]/[3760]	Speed: 69.698875 samples/sec	 accuracy=53.631602	 loss=1.962311	 lr=0.010000
Epoch[030] Batch [3599]/[3760]	Speed: 69.733917 samples/sec	 accuracy=53.627170	 loss=1.962670	 lr=0.010000
Epoch[030] Batch [3649]/[3760]	Speed: 68.845083 samples/sec	 accuracy=53.606592	 loss=1.963514	 lr=0.010000
Epoch[030] Batch [3699]/[3760]	Speed: 67.973570 samples/sec	 accuracy=53.598818	 loss=1.964558	 lr=0.010000
Epoch[030] Batch [3749]/[3760]	Speed: 78.006828 samples/sec	 accuracy=53.597917	 loss=1.964887	 lr=0.010000
Batch [0049]/[0303]: acc-top1=54.718750 acc-top5=77.937500
Batch [0099]/[0303]: acc-top1=53.750000 acc-top5=78.015625
Batch [0149]/[0303]: acc-top1=54.000000 acc-top5=78.552083
Batch [0199]/[0303]: acc-top1=54.257812 acc-top5=78.828125
Batch [0249]/[0303]: acc-top1=54.318750 acc-top5=78.762500
Batch [0299]/[0303]: acc-top1=54.072917 acc-top5=78.510417
[Epoch 030] training: accuracy=53.597906	 loss=1.964972
[Epoch 030] speed: 69 samples/sec	time cost: 3767.071199
[Epoch 030] validation: acc-top1=54.058375 acc-top5=78.485974 loss=2.046966
Epoch[031] Batch [0049]/[3760]	Speed: 48.190521 samples/sec	 accuracy=55.343750	 loss=1.860764	 lr=0.010000
Epoch[031] Batch [0099]/[3760]	Speed: 68.346049 samples/sec	 accuracy=54.796875	 loss=1.919288	 lr=0.010000
Epoch[031] Batch [0149]/[3760]	Speed: 67.707063 samples/sec	 accuracy=54.781250	 loss=1.916808	 lr=0.010000
Epoch[031] Batch [0199]/[3760]	Speed: 70.688704 samples/sec	 accuracy=54.710938	 loss=1.911616	 lr=0.010000
Epoch[031] Batch [0249]/[3760]	Speed: 70.674853 samples/sec	 accuracy=54.750000	 loss=1.909824	 lr=0.010000
Epoch[031] Batch [0299]/[3760]	Speed: 69.468505 samples/sec	 accuracy=54.744792	 loss=1.909177	 lr=0.010000
Epoch[031] Batch [0349]/[3760]	Speed: 69.877602 samples/sec	 accuracy=54.709821	 loss=1.908712	 lr=0.010000
Epoch[031] Batch [0399]/[3760]	Speed: 70.099176 samples/sec	 accuracy=54.761719	 loss=1.908078	 lr=0.010000
Epoch[031] Batch [0449]/[3760]	Speed: 69.317793 samples/sec	 accuracy=54.677083	 loss=1.912687	 lr=0.010000
Epoch[031] Batch [0499]/[3760]	Speed: 70.114503 samples/sec	 accuracy=54.559375	 loss=1.918832	 lr=0.010000
Epoch[031] Batch [0549]/[3760]	Speed: 69.990064 samples/sec	 accuracy=54.446023	 loss=1.924803	 lr=0.010000
Epoch[031] Batch [0599]/[3760]	Speed: 68.441540 samples/sec	 accuracy=54.385417	 loss=1.926181	 lr=0.010000
Epoch[031] Batch [0649]/[3760]	Speed: 69.584189 samples/sec	 accuracy=54.281250	 loss=1.927494	 lr=0.010000
Epoch[031] Batch [0699]/[3760]	Speed: 69.717652 samples/sec	 accuracy=54.272321	 loss=1.926384	 lr=0.010000
Epoch[031] Batch [0749]/[3760]	Speed: 70.220538 samples/sec	 accuracy=54.193750	 loss=1.926738	 lr=0.010000
Epoch[031] Batch [0799]/[3760]	Speed: 68.268615 samples/sec	 accuracy=54.117188	 loss=1.931786	 lr=0.010000
Epoch[031] Batch [0849]/[3760]	Speed: 70.324644 samples/sec	 accuracy=54.101103	 loss=1.932701	 lr=0.010000
Epoch[031] Batch [0899]/[3760]	Speed: 70.625720 samples/sec	 accuracy=54.109375	 loss=1.933716	 lr=0.010000
Epoch[031] Batch [0949]/[3760]	Speed: 69.096695 samples/sec	 accuracy=54.097039	 loss=1.933812	 lr=0.010000
Epoch[031] Batch [0999]/[3760]	Speed: 69.295446 samples/sec	 accuracy=54.059375	 loss=1.933218	 lr=0.010000
Epoch[031] Batch [1049]/[3760]	Speed: 70.177894 samples/sec	 accuracy=54.040179	 loss=1.935019	 lr=0.010000
Epoch[031] Batch [1099]/[3760]	Speed: 69.792133 samples/sec	 accuracy=54.035511	 loss=1.934703	 lr=0.010000
Epoch[031] Batch [1149]/[3760]	Speed: 69.216926 samples/sec	 accuracy=54.043478	 loss=1.936015	 lr=0.010000
Epoch[031] Batch [1199]/[3760]	Speed: 69.371925 samples/sec	 accuracy=54.035156	 loss=1.937302	 lr=0.010000
Epoch[031] Batch [1249]/[3760]	Speed: 69.792183 samples/sec	 accuracy=53.975000	 loss=1.939269	 lr=0.010000
Epoch[031] Batch [1299]/[3760]	Speed: 69.794566 samples/sec	 accuracy=53.993990	 loss=1.940033	 lr=0.010000
Epoch[031] Batch [1349]/[3760]	Speed: 69.785227 samples/sec	 accuracy=53.994213	 loss=1.939422	 lr=0.010000
Epoch[031] Batch [1399]/[3760]	Speed: 70.349993 samples/sec	 accuracy=53.955357	 loss=1.940499	 lr=0.010000
Epoch[031] Batch [1449]/[3760]	Speed: 67.481667 samples/sec	 accuracy=53.966595	 loss=1.940601	 lr=0.010000
Epoch[031] Batch [1499]/[3760]	Speed: 70.738888 samples/sec	 accuracy=53.959375	 loss=1.940514	 lr=0.010000
Epoch[031] Batch [1549]/[3760]	Speed: 71.101238 samples/sec	 accuracy=53.934476	 loss=1.941200	 lr=0.010000
Epoch[031] Batch [1599]/[3760]	Speed: 70.299825 samples/sec	 accuracy=53.901367	 loss=1.942439	 lr=0.010000
Epoch[031] Batch [1649]/[3760]	Speed: 68.191286 samples/sec	 accuracy=53.897727	 loss=1.942955	 lr=0.010000
Epoch[031] Batch [1699]/[3760]	Speed: 70.780907 samples/sec	 accuracy=53.941176	 loss=1.941393	 lr=0.010000
Epoch[031] Batch [1749]/[3760]	Speed: 67.184993 samples/sec	 accuracy=53.907143	 loss=1.942662	 lr=0.010000
Epoch[031] Batch [1799]/[3760]	Speed: 70.824951 samples/sec	 accuracy=53.886285	 loss=1.942694	 lr=0.010000
Epoch[031] Batch [1849]/[3760]	Speed: 70.970889 samples/sec	 accuracy=53.889358	 loss=1.942470	 lr=0.010000
Epoch[031] Batch [1899]/[3760]	Speed: 69.895441 samples/sec	 accuracy=53.917763	 loss=1.942104	 lr=0.010000
Epoch[031] Batch [1949]/[3760]	Speed: 69.873289 samples/sec	 accuracy=53.927083	 loss=1.941727	 lr=0.010000
Epoch[031] Batch [1999]/[3760]	Speed: 70.259148 samples/sec	 accuracy=53.911719	 loss=1.941629	 lr=0.010000
Epoch[031] Batch [2049]/[3760]	Speed: 70.058068 samples/sec	 accuracy=53.912348	 loss=1.942211	 lr=0.010000
Epoch[031] Batch [2099]/[3760]	Speed: 69.108528 samples/sec	 accuracy=53.883185	 loss=1.944778	 lr=0.010000
Epoch[031] Batch [2149]/[3760]	Speed: 68.835329 samples/sec	 accuracy=53.845930	 loss=1.946823	 lr=0.010000
Epoch[031] Batch [2199]/[3760]	Speed: 70.857201 samples/sec	 accuracy=53.844460	 loss=1.946212	 lr=0.010000
Epoch[031] Batch [2249]/[3760]	Speed: 70.714937 samples/sec	 accuracy=53.835417	 loss=1.946702	 lr=0.010000
Epoch[031] Batch [2299]/[3760]	Speed: 69.830801 samples/sec	 accuracy=53.790082	 loss=1.949232	 lr=0.010000
Epoch[031] Batch [2349]/[3760]	Speed: 69.716900 samples/sec	 accuracy=53.754654	 loss=1.951610	 lr=0.010000
Epoch[031] Batch [2399]/[3760]	Speed: 70.428327 samples/sec	 accuracy=53.745443	 loss=1.952187	 lr=0.010000
Epoch[031] Batch [2449]/[3760]	Speed: 70.379014 samples/sec	 accuracy=53.723214	 loss=1.953157	 lr=0.010000
Epoch[031] Batch [2499]/[3760]	Speed: 69.087850 samples/sec	 accuracy=53.724375	 loss=1.953265	 lr=0.010000
Epoch[031] Batch [2549]/[3760]	Speed: 70.475244 samples/sec	 accuracy=53.744485	 loss=1.953018	 lr=0.010000
Epoch[031] Batch [2599]/[3760]	Speed: 69.521783 samples/sec	 accuracy=53.753005	 loss=1.952669	 lr=0.010000
Epoch[031] Batch [2649]/[3760]	Speed: 69.892854 samples/sec	 accuracy=53.774175	 loss=1.952385	 lr=0.010000
Epoch[031] Batch [2699]/[3760]	Speed: 69.968990 samples/sec	 accuracy=53.752315	 loss=1.953348	 lr=0.010000
Epoch[031] Batch [2749]/[3760]	Speed: 69.686652 samples/sec	 accuracy=53.742045	 loss=1.954066	 lr=0.010000
Epoch[031] Batch [2799]/[3760]	Speed: 69.178470 samples/sec	 accuracy=53.733817	 loss=1.954657	 lr=0.010000
Epoch[031] Batch [2849]/[3760]	Speed: 70.057242 samples/sec	 accuracy=53.750000	 loss=1.954555	 lr=0.010000
Epoch[031] Batch [2899]/[3760]	Speed: 69.887827 samples/sec	 accuracy=53.760776	 loss=1.954075	 lr=0.010000
Epoch[031] Batch [2949]/[3760]	Speed: 69.841043 samples/sec	 accuracy=53.766949	 loss=1.954234	 lr=0.010000
Epoch[031] Batch [2999]/[3760]	Speed: 70.057396 samples/sec	 accuracy=53.740625	 loss=1.955130	 lr=0.010000
Epoch[031] Batch [3049]/[3760]	Speed: 69.911460 samples/sec	 accuracy=53.736680	 loss=1.955817	 lr=0.010000
Epoch[031] Batch [3099]/[3760]	Speed: 68.771148 samples/sec	 accuracy=53.732863	 loss=1.955928	 lr=0.010000
Epoch[031] Batch [3149]/[3760]	Speed: 70.314494 samples/sec	 accuracy=53.737599	 loss=1.956100	 lr=0.010000
Epoch[031] Batch [3199]/[3760]	Speed: 69.743853 samples/sec	 accuracy=53.732910	 loss=1.956853	 lr=0.010000
Epoch[031] Batch [3249]/[3760]	Speed: 69.802929 samples/sec	 accuracy=53.726442	 loss=1.956512	 lr=0.010000
Epoch[031] Batch [3299]/[3760]	Speed: 70.620481 samples/sec	 accuracy=53.732955	 loss=1.956038	 lr=0.010000
Epoch[031] Batch [3349]/[3760]	Speed: 68.900277 samples/sec	 accuracy=53.747201	 loss=1.955622	 lr=0.010000
Epoch[031] Batch [3399]/[3760]	Speed: 68.574150 samples/sec	 accuracy=53.739890	 loss=1.955786	 lr=0.010000
Epoch[031] Batch [3449]/[3760]	Speed: 71.003621 samples/sec	 accuracy=53.758605	 loss=1.954980	 lr=0.010000
Epoch[031] Batch [3499]/[3760]	Speed: 69.994735 samples/sec	 accuracy=53.776339	 loss=1.954843	 lr=0.010000
Epoch[031] Batch [3549]/[3760]	Speed: 71.023556 samples/sec	 accuracy=53.768046	 loss=1.955482	 lr=0.010000
Epoch[031] Batch [3599]/[3760]	Speed: 68.823681 samples/sec	 accuracy=53.744358	 loss=1.956703	 lr=0.010000
Epoch[031] Batch [3649]/[3760]	Speed: 69.985877 samples/sec	 accuracy=53.734589	 loss=1.956901	 lr=0.010000
Epoch[031] Batch [3699]/[3760]	Speed: 69.876485 samples/sec	 accuracy=53.730574	 loss=1.957371	 lr=0.010000
Epoch[031] Batch [3749]/[3760]	Speed: 77.198420 samples/sec	 accuracy=53.718750	 loss=1.957420	 lr=0.010000
Batch [0049]/[0303]: acc-top1=53.031250 acc-top5=78.531250
Batch [0099]/[0303]: acc-top1=52.625000 acc-top5=78.093750
Batch [0149]/[0303]: acc-top1=53.479167 acc-top5=78.708333
Batch [0199]/[0303]: acc-top1=54.054688 acc-top5=78.984375
Batch [0249]/[0303]: acc-top1=54.018750 acc-top5=79.262500
Batch [0299]/[0303]: acc-top1=53.890625 acc-top5=79.000000
[Epoch 031] training: accuracy=53.725066	 loss=1.957167
[Epoch 031] speed: 69 samples/sec	time cost: 3756.870546
[Epoch 031] validation: acc-top1=53.846947 acc-top5=78.996493 loss=2.039789
Epoch[032] Batch [0049]/[3759]	Speed: 48.853311 samples/sec	 accuracy=54.500000	 loss=1.918217	 lr=0.010000
Epoch[032] Batch [0099]/[3759]	Speed: 67.838768 samples/sec	 accuracy=54.078125	 loss=1.953114	 lr=0.010000
Epoch[032] Batch [0149]/[3759]	Speed: 69.746590 samples/sec	 accuracy=53.406250	 loss=1.957096	 lr=0.010000
Epoch[032] Batch [0199]/[3759]	Speed: 69.460610 samples/sec	 accuracy=53.484375	 loss=1.947059	 lr=0.010000
Epoch[032] Batch [0249]/[3759]	Speed: 69.769594 samples/sec	 accuracy=53.850000	 loss=1.940099	 lr=0.010000
Epoch[032] Batch [0299]/[3759]	Speed: 69.070901 samples/sec	 accuracy=54.098958	 loss=1.931165	 lr=0.010000
Epoch[032] Batch [0349]/[3759]	Speed: 69.469974 samples/sec	 accuracy=54.129464	 loss=1.929368	 lr=0.010000
Epoch[032] Batch [0399]/[3759]	Speed: 70.904285 samples/sec	 accuracy=54.082031	 loss=1.933496	 lr=0.010000
Epoch[032] Batch [0449]/[3759]	Speed: 68.900717 samples/sec	 accuracy=54.024306	 loss=1.933229	 lr=0.010000
Epoch[032] Batch [0499]/[3759]	Speed: 70.421113 samples/sec	 accuracy=54.046875	 loss=1.929383	 lr=0.010000
Epoch[032] Batch [0549]/[3759]	Speed: 69.088331 samples/sec	 accuracy=53.997159	 loss=1.931164	 lr=0.010000
Epoch[032] Batch [0599]/[3759]	Speed: 70.578746 samples/sec	 accuracy=54.018229	 loss=1.931911	 lr=0.010000
Epoch[032] Batch [0649]/[3759]	Speed: 69.617010 samples/sec	 accuracy=53.911058	 loss=1.936206	 lr=0.010000
Epoch[032] Batch [0699]/[3759]	Speed: 68.370194 samples/sec	 accuracy=53.897321	 loss=1.937778	 lr=0.010000
Epoch[032] Batch [0749]/[3759]	Speed: 70.720364 samples/sec	 accuracy=53.910417	 loss=1.935691	 lr=0.010000
Epoch[032] Batch [0799]/[3759]	Speed: 70.368952 samples/sec	 accuracy=53.812500	 loss=1.940717	 lr=0.010000
Epoch[032] Batch [0849]/[3759]	Speed: 70.130766 samples/sec	 accuracy=53.731618	 loss=1.944047	 lr=0.010000
Epoch[032] Batch [0899]/[3759]	Speed: 70.552496 samples/sec	 accuracy=53.744792	 loss=1.944681	 lr=0.010000
Epoch[032] Batch [0949]/[3759]	Speed: 67.696766 samples/sec	 accuracy=53.722039	 loss=1.944657	 lr=0.010000
Epoch[032] Batch [0999]/[3759]	Speed: 70.086301 samples/sec	 accuracy=53.659375	 loss=1.945034	 lr=0.010000
Epoch[032] Batch [1049]/[3759]	Speed: 70.212043 samples/sec	 accuracy=53.623512	 loss=1.947872	 lr=0.010000
Epoch[032] Batch [1099]/[3759]	Speed: 70.414704 samples/sec	 accuracy=53.734375	 loss=1.944232	 lr=0.010000
Epoch[032] Batch [1149]/[3759]	Speed: 69.760280 samples/sec	 accuracy=53.707880	 loss=1.946960	 lr=0.010000
Epoch[032] Batch [1199]/[3759]	Speed: 70.192438 samples/sec	 accuracy=53.755208	 loss=1.945173	 lr=0.010000
Epoch[032] Batch [1249]/[3759]	Speed: 70.104477 samples/sec	 accuracy=53.820000	 loss=1.944425	 lr=0.010000
Epoch[032] Batch [1299]/[3759]	Speed: 68.351684 samples/sec	 accuracy=53.849760	 loss=1.945389	 lr=0.010000
Epoch[032] Batch [1349]/[3759]	Speed: 69.288461 samples/sec	 accuracy=53.868056	 loss=1.945332	 lr=0.010000
Epoch[032] Batch [1399]/[3759]	Speed: 69.663122 samples/sec	 accuracy=53.849330	 loss=1.946599	 lr=0.010000
Epoch[032] Batch [1449]/[3759]	Speed: 70.951324 samples/sec	 accuracy=53.826509	 loss=1.946529	 lr=0.010000
Epoch[032] Batch [1499]/[3759]	Speed: 69.143218 samples/sec	 accuracy=53.822917	 loss=1.947872	 lr=0.010000
Epoch[032] Batch [1549]/[3759]	Speed: 70.902499 samples/sec	 accuracy=53.867944	 loss=1.946440	 lr=0.010000
Epoch[032] Batch [1599]/[3759]	Speed: 69.598257 samples/sec	 accuracy=53.851563	 loss=1.947955	 lr=0.010000
Epoch[032] Batch [1649]/[3759]	Speed: 68.080468 samples/sec	 accuracy=53.867424	 loss=1.949063	 lr=0.010000
Epoch[032] Batch [1699]/[3759]	Speed: 69.472217 samples/sec	 accuracy=53.867647	 loss=1.949904	 lr=0.010000
Epoch[032] Batch [1749]/[3759]	Speed: 69.282620 samples/sec	 accuracy=53.875000	 loss=1.949348	 lr=0.010000
Epoch[032] Batch [1799]/[3759]	Speed: 71.274383 samples/sec	 accuracy=53.875000	 loss=1.950094	 lr=0.010000
Epoch[032] Batch [1849]/[3759]	Speed: 69.725937 samples/sec	 accuracy=53.880068	 loss=1.950488	 lr=0.010000
Epoch[032] Batch [1899]/[3759]	Speed: 69.274410 samples/sec	 accuracy=53.904605	 loss=1.950211	 lr=0.010000
Epoch[032] Batch [1949]/[3759]	Speed: 70.185939 samples/sec	 accuracy=53.906250	 loss=1.950472	 lr=0.010000
Epoch[032] Batch [1999]/[3759]	Speed: 69.422916 samples/sec	 accuracy=53.903125	 loss=1.951198	 lr=0.010000
Epoch[032] Batch [2049]/[3759]	Speed: 69.817707 samples/sec	 accuracy=53.964177	 loss=1.949025	 lr=0.010000
Epoch[032] Batch [2099]/[3759]	Speed: 68.700866 samples/sec	 accuracy=53.958333	 loss=1.948320	 lr=0.010000
Epoch[032] Batch [2149]/[3759]	Speed: 70.992716 samples/sec	 accuracy=53.943314	 loss=1.949353	 lr=0.010000
Epoch[032] Batch [2199]/[3759]	Speed: 69.808093 samples/sec	 accuracy=53.942472	 loss=1.949389	 lr=0.010000
Epoch[032] Batch [2249]/[3759]	Speed: 69.709245 samples/sec	 accuracy=53.922917	 loss=1.950348	 lr=0.010000
Epoch[032] Batch [2299]/[3759]	Speed: 70.447850 samples/sec	 accuracy=53.902174	 loss=1.951360	 lr=0.010000
Epoch[032] Batch [2349]/[3759]	Speed: 68.845816 samples/sec	 accuracy=53.908245	 loss=1.950661	 lr=0.010000
Epoch[032] Batch [2399]/[3759]	Speed: 70.700534 samples/sec	 accuracy=53.903646	 loss=1.951885	 lr=0.010000
Epoch[032] Batch [2449]/[3759]	Speed: 69.060649 samples/sec	 accuracy=53.899872	 loss=1.952276	 lr=0.010000
Epoch[032] Batch [2499]/[3759]	Speed: 71.448614 samples/sec	 accuracy=53.892500	 loss=1.952402	 lr=0.010000
Epoch[032] Batch [2549]/[3759]	Speed: 68.884679 samples/sec	 accuracy=53.876225	 loss=1.953098	 lr=0.010000
Epoch[032] Batch [2599]/[3759]	Speed: 69.795072 samples/sec	 accuracy=53.861779	 loss=1.954180	 lr=0.010000
Epoch[032] Batch [2649]/[3759]	Speed: 69.294119 samples/sec	 accuracy=53.881486	 loss=1.953634	 lr=0.010000
Epoch[032] Batch [2699]/[3759]	Speed: 70.396535 samples/sec	 accuracy=53.863426	 loss=1.954394	 lr=0.010000
Epoch[032] Batch [2749]/[3759]	Speed: 70.187503 samples/sec	 accuracy=53.890909	 loss=1.953591	 lr=0.010000
Epoch[032] Batch [2799]/[3759]	Speed: 70.597401 samples/sec	 accuracy=53.854911	 loss=1.954282	 lr=0.010000
Epoch[032] Batch [2849]/[3759]	Speed: 68.235827 samples/sec	 accuracy=53.859101	 loss=1.954137	 lr=0.010000
Epoch[032] Batch [2899]/[3759]	Speed: 67.612492 samples/sec	 accuracy=53.842672	 loss=1.954691	 lr=0.010000
Epoch[032] Batch [2949]/[3759]	Speed: 69.545201 samples/sec	 accuracy=53.838453	 loss=1.954801	 lr=0.010000
Epoch[032] Batch [2999]/[3759]	Speed: 72.220008 samples/sec	 accuracy=53.826042	 loss=1.955023	 lr=0.010000
Epoch[032] Batch [3049]/[3759]	Speed: 67.780271 samples/sec	 accuracy=53.826332	 loss=1.955363	 lr=0.010000
Epoch[032] Batch [3099]/[3759]	Speed: 71.256264 samples/sec	 accuracy=53.836694	 loss=1.954751	 lr=0.010000
Epoch[032] Batch [3149]/[3759]	Speed: 68.700449 samples/sec	 accuracy=53.837302	 loss=1.955199	 lr=0.010000
Epoch[032] Batch [3199]/[3759]	Speed: 69.689354 samples/sec	 accuracy=53.846191	 loss=1.955325	 lr=0.010000
Epoch[032] Batch [3249]/[3759]	Speed: 69.144461 samples/sec	 accuracy=53.851442	 loss=1.955061	 lr=0.010000
Epoch[032] Batch [3299]/[3759]	Speed: 71.107097 samples/sec	 accuracy=53.864110	 loss=1.954593	 lr=0.010000
Epoch[032] Batch [3349]/[3759]	Speed: 66.863502 samples/sec	 accuracy=53.826959	 loss=1.956160	 lr=0.010000
Epoch[032] Batch [3399]/[3759]	Speed: 70.681869 samples/sec	 accuracy=53.842831	 loss=1.956357	 lr=0.010000
Epoch[032] Batch [3449]/[3759]	Speed: 68.606779 samples/sec	 accuracy=53.834239	 loss=1.956682	 lr=0.010000
Epoch[032] Batch [3499]/[3759]	Speed: 71.273619 samples/sec	 accuracy=53.824107	 loss=1.957019	 lr=0.010000
Epoch[032] Batch [3549]/[3759]	Speed: 70.628988 samples/sec	 accuracy=53.813380	 loss=1.957431	 lr=0.010000
Epoch[032] Batch [3599]/[3759]	Speed: 67.126417 samples/sec	 accuracy=53.800347	 loss=1.958216	 lr=0.010000
Epoch[032] Batch [3649]/[3759]	Speed: 71.303110 samples/sec	 accuracy=53.786815	 loss=1.958990	 lr=0.010000
Epoch[032] Batch [3699]/[3759]	Speed: 70.634469 samples/sec	 accuracy=53.776605	 loss=1.959365	 lr=0.010000
Epoch[032] Batch [3749]/[3759]	Speed: 77.537692 samples/sec	 accuracy=53.788333	 loss=1.958394	 lr=0.010000
Batch [0049]/[0303]: acc-top1=53.906250 acc-top5=79.000000
Batch [0099]/[0303]: acc-top1=54.218750 acc-top5=78.640625
Batch [0149]/[0303]: acc-top1=55.052083 acc-top5=79.302083
Batch [0199]/[0303]: acc-top1=55.726562 acc-top5=79.562500
Batch [0249]/[0303]: acc-top1=55.731250 acc-top5=79.700000
Batch [0299]/[0303]: acc-top1=55.453125 acc-top5=79.520833
[Epoch 032] training: accuracy=53.786745	 loss=1.958506
[Epoch 032] speed: 69 samples/sec	time cost: 3758.098127
[Epoch 032] validation: acc-top1=55.383663 acc-top5=79.445132 loss=1.985938
Epoch[033] Batch [0049]/[3760]	Speed: 48.574225 samples/sec	 accuracy=55.500000	 loss=1.857870	 lr=0.010000
Epoch[033] Batch [0099]/[3760]	Speed: 68.269300 samples/sec	 accuracy=55.421875	 loss=1.868860	 lr=0.010000
Epoch[033] Batch [0149]/[3760]	Speed: 69.775138 samples/sec	 accuracy=55.000000	 loss=1.889954	 lr=0.010000
Epoch[033] Batch [0199]/[3760]	Speed: 68.828694 samples/sec	 accuracy=54.578125	 loss=1.910561	 lr=0.010000
Epoch[033] Batch [0249]/[3760]	Speed: 70.920135 samples/sec	 accuracy=54.475000	 loss=1.921620	 lr=0.010000
Epoch[033] Batch [0299]/[3760]	Speed: 68.790152 samples/sec	 accuracy=54.343750	 loss=1.923559	 lr=0.010000
Epoch[033] Batch [0349]/[3760]	Speed: 68.356306 samples/sec	 accuracy=54.392857	 loss=1.920978	 lr=0.010000
Epoch[033] Batch [0399]/[3760]	Speed: 70.799025 samples/sec	 accuracy=54.363281	 loss=1.922942	 lr=0.010000
Epoch[033] Batch [0449]/[3760]	Speed: 69.605811 samples/sec	 accuracy=54.281250	 loss=1.922890	 lr=0.010000
Epoch[033] Batch [0499]/[3760]	Speed: 70.895923 samples/sec	 accuracy=54.312500	 loss=1.923570	 lr=0.010000
Epoch[033] Batch [0549]/[3760]	Speed: 68.005463 samples/sec	 accuracy=54.389205	 loss=1.920937	 lr=0.010000
Epoch[033] Batch [0599]/[3760]	Speed: 70.657539 samples/sec	 accuracy=54.322917	 loss=1.924667	 lr=0.010000
Epoch[033] Batch [0649]/[3760]	Speed: 68.917012 samples/sec	 accuracy=54.262019	 loss=1.928631	 lr=0.010000
Epoch[033] Batch [0699]/[3760]	Speed: 71.619242 samples/sec	 accuracy=54.229911	 loss=1.929145	 lr=0.010000
Epoch[033] Batch [0749]/[3760]	Speed: 66.595730 samples/sec	 accuracy=54.135417	 loss=1.931595	 lr=0.010000
Epoch[033] Batch [0799]/[3760]	Speed: 71.354382 samples/sec	 accuracy=54.076172	 loss=1.929813	 lr=0.010000
Epoch[033] Batch [0849]/[3760]	Speed: 70.123444 samples/sec	 accuracy=53.941176	 loss=1.932839	 lr=0.010000
Epoch[033] Batch [0899]/[3760]	Speed: 70.176621 samples/sec	 accuracy=53.918403	 loss=1.932282	 lr=0.010000
Epoch[033] Batch [0949]/[3760]	Speed: 68.411975 samples/sec	 accuracy=53.937500	 loss=1.935193	 lr=0.010000
Epoch[033] Batch [0999]/[3760]	Speed: 69.330429 samples/sec	 accuracy=54.014063	 loss=1.933203	 lr=0.010000
Epoch[033] Batch [1049]/[3760]	Speed: 68.948466 samples/sec	 accuracy=54.055060	 loss=1.930360	 lr=0.010000
Epoch[033] Batch [1099]/[3760]	Speed: 69.501381 samples/sec	 accuracy=54.022727	 loss=1.931151	 lr=0.010000
Epoch[033] Batch [1149]/[3760]	Speed: 69.333928 samples/sec	 accuracy=54.036685	 loss=1.932041	 lr=0.010000
Epoch[033] Batch [1199]/[3760]	Speed: 70.656448 samples/sec	 accuracy=54.072917	 loss=1.930617	 lr=0.010000
Epoch[033] Batch [1249]/[3760]	Speed: 70.034168 samples/sec	 accuracy=54.056250	 loss=1.931670	 lr=0.010000
Epoch[033] Batch [1299]/[3760]	Speed: 66.291296 samples/sec	 accuracy=54.037260	 loss=1.931167	 lr=0.010000
Epoch[033] Batch [1349]/[3760]	Speed: 69.517982 samples/sec	 accuracy=54.067130	 loss=1.930619	 lr=0.010000
Epoch[033] Batch [1399]/[3760]	Speed: 71.562329 samples/sec	 accuracy=54.085938	 loss=1.931004	 lr=0.010000
Epoch[033] Batch [1449]/[3760]	Speed: 68.540412 samples/sec	 accuracy=54.075431	 loss=1.931160	 lr=0.010000
Epoch[033] Batch [1499]/[3760]	Speed: 70.820642 samples/sec	 accuracy=54.076042	 loss=1.931694	 lr=0.010000
Epoch[033] Batch [1549]/[3760]	Speed: 70.026356 samples/sec	 accuracy=54.059476	 loss=1.933276	 lr=0.010000
Epoch[033] Batch [1599]/[3760]	Speed: 70.113568 samples/sec	 accuracy=54.002930	 loss=1.935407	 lr=0.010000
Epoch[033] Batch [1649]/[3760]	Speed: 70.008190 samples/sec	 accuracy=54.001894	 loss=1.935370	 lr=0.010000
Epoch[033] Batch [1699]/[3760]	Speed: 68.791196 samples/sec	 accuracy=53.963235	 loss=1.935839	 lr=0.010000
Epoch[033] Batch [1749]/[3760]	Speed: 70.581630 samples/sec	 accuracy=53.936607	 loss=1.936707	 lr=0.010000
Epoch[033] Batch [1799]/[3760]	Speed: 67.940837 samples/sec	 accuracy=53.943576	 loss=1.936704	 lr=0.010000
Epoch[033] Batch [1849]/[3760]	Speed: 70.665754 samples/sec	 accuracy=53.947635	 loss=1.937763	 lr=0.010000
Epoch[033] Batch [1899]/[3760]	Speed: 69.329752 samples/sec	 accuracy=53.943257	 loss=1.937364	 lr=0.010000
Epoch[033] Batch [1949]/[3760]	Speed: 69.232900 samples/sec	 accuracy=53.957532	 loss=1.937498	 lr=0.010000
Epoch[033] Batch [1999]/[3760]	Speed: 69.436587 samples/sec	 accuracy=53.951563	 loss=1.937689	 lr=0.010000
Epoch[033] Batch [2049]/[3760]	Speed: 68.637356 samples/sec	 accuracy=53.961128	 loss=1.936879	 lr=0.010000
Epoch[033] Batch [2099]/[3760]	Speed: 71.206686 samples/sec	 accuracy=53.960565	 loss=1.936698	 lr=0.010000
Epoch[033] Batch [2149]/[3760]	Speed: 68.517919 samples/sec	 accuracy=53.941860	 loss=1.936822	 lr=0.010000
Epoch[033] Batch [2199]/[3760]	Speed: 69.818603 samples/sec	 accuracy=53.945312	 loss=1.937304	 lr=0.010000
Epoch[033] Batch [2249]/[3760]	Speed: 69.324195 samples/sec	 accuracy=53.976389	 loss=1.936788	 lr=0.010000
Epoch[033] Batch [2299]/[3760]	Speed: 69.586510 samples/sec	 accuracy=53.906929	 loss=1.939480	 lr=0.010000
Epoch[033] Batch [2349]/[3760]	Speed: 70.305818 samples/sec	 accuracy=53.948138	 loss=1.937869	 lr=0.010000
Epoch[033] Batch [2399]/[3760]	Speed: 69.307153 samples/sec	 accuracy=53.969401	 loss=1.937172	 lr=0.010000
Epoch[033] Batch [2449]/[3760]	Speed: 67.789345 samples/sec	 accuracy=53.970026	 loss=1.937398	 lr=0.010000
Epoch[033] Batch [2499]/[3760]	Speed: 69.991876 samples/sec	 accuracy=53.983125	 loss=1.937598	 lr=0.010000
Epoch[033] Batch [2549]/[3760]	Speed: 71.471105 samples/sec	 accuracy=53.990196	 loss=1.937350	 lr=0.010000
Epoch[033] Batch [2599]/[3760]	Speed: 69.162569 samples/sec	 accuracy=53.966947	 loss=1.938153	 lr=0.010000
Epoch[033] Batch [2649]/[3760]	Speed: 69.536587 samples/sec	 accuracy=53.977594	 loss=1.937951	 lr=0.010000
Epoch[033] Batch [2699]/[3760]	Speed: 71.277609 samples/sec	 accuracy=53.992477	 loss=1.938054	 lr=0.010000
Epoch[033] Batch [2749]/[3760]	Speed: 68.647310 samples/sec	 accuracy=53.985227	 loss=1.937898	 lr=0.010000
Epoch[033] Batch [2799]/[3760]	Speed: 69.108165 samples/sec	 accuracy=53.995536	 loss=1.937948	 lr=0.010000
Epoch[033] Batch [2849]/[3760]	Speed: 69.542858 samples/sec	 accuracy=54.010417	 loss=1.938069	 lr=0.010000
Epoch[033] Batch [2899]/[3760]	Speed: 69.664683 samples/sec	 accuracy=54.019935	 loss=1.938394	 lr=0.010000
Epoch[033] Batch [2949]/[3760]	Speed: 70.115655 samples/sec	 accuracy=54.030191	 loss=1.938573	 lr=0.010000
Epoch[033] Batch [2999]/[3760]	Speed: 69.380639 samples/sec	 accuracy=54.043229	 loss=1.938788	 lr=0.010000
Epoch[033] Batch [3049]/[3760]	Speed: 69.089172 samples/sec	 accuracy=54.056865	 loss=1.938721	 lr=0.010000
Epoch[033] Batch [3099]/[3760]	Speed: 70.662108 samples/sec	 accuracy=54.028226	 loss=1.940136	 lr=0.010000
Epoch[033] Batch [3149]/[3760]	Speed: 69.464257 samples/sec	 accuracy=54.026290	 loss=1.939929	 lr=0.010000
Epoch[033] Batch [3199]/[3760]	Speed: 70.511357 samples/sec	 accuracy=54.024414	 loss=1.940779	 lr=0.010000
Epoch[033] Batch [3249]/[3760]	Speed: 68.287895 samples/sec	 accuracy=54.009135	 loss=1.941181	 lr=0.010000
Epoch[033] Batch [3299]/[3760]	Speed: 70.326653 samples/sec	 accuracy=54.014678	 loss=1.941048	 lr=0.010000
Epoch[033] Batch [3349]/[3760]	Speed: 70.401963 samples/sec	 accuracy=54.005597	 loss=1.941232	 lr=0.010000
Epoch[033] Batch [3399]/[3760]	Speed: 70.549845 samples/sec	 accuracy=54.006434	 loss=1.941076	 lr=0.010000
Epoch[033] Batch [3449]/[3760]	Speed: 68.788167 samples/sec	 accuracy=54.009058	 loss=1.941263	 lr=0.010000
Epoch[033] Batch [3499]/[3760]	Speed: 69.728773 samples/sec	 accuracy=53.978125	 loss=1.941988	 lr=0.010000
Epoch[033] Batch [3549]/[3760]	Speed: 68.699892 samples/sec	 accuracy=53.957746	 loss=1.942892	 lr=0.010000
Epoch[033] Batch [3599]/[3760]	Speed: 70.954037 samples/sec	 accuracy=53.958767	 loss=1.942496	 lr=0.010000
Epoch[033] Batch [3649]/[3760]	Speed: 70.459318 samples/sec	 accuracy=53.938784	 loss=1.943471	 lr=0.010000
Epoch[033] Batch [3699]/[3760]	Speed: 69.648617 samples/sec	 accuracy=53.941301	 loss=1.943506	 lr=0.010000
Epoch[033] Batch [3749]/[3760]	Speed: 76.744700 samples/sec	 accuracy=53.941250	 loss=1.943914	 lr=0.010000
Batch [0049]/[0303]: acc-top1=54.093750 acc-top5=78.906250
Batch [0099]/[0303]: acc-top1=53.796875 acc-top5=78.625000
Batch [0149]/[0303]: acc-top1=54.375000 acc-top5=79.125000
Batch [0199]/[0303]: acc-top1=54.875000 acc-top5=79.296875
Batch [0249]/[0303]: acc-top1=54.868750 acc-top5=79.475000
Batch [0299]/[0303]: acc-top1=54.604167 acc-top5=79.156250
[Epoch 033] training: accuracy=53.941988	 loss=1.943971
[Epoch 033] speed: 69 samples/sec	time cost: 3763.992796
[Epoch 033] validation: acc-top1=54.579208 acc-top5=79.146040 loss=1.998207
Epoch[034] Batch [0049]/[3759]	Speed: 47.792370 samples/sec	 accuracy=53.843750	 loss=1.912849	 lr=0.010000
Epoch[034] Batch [0099]/[3759]	Speed: 69.053457 samples/sec	 accuracy=54.921875	 loss=1.891521	 lr=0.010000
Epoch[034] Batch [0149]/[3759]	Speed: 70.097442 samples/sec	 accuracy=54.968750	 loss=1.882895	 lr=0.010000
Epoch[034] Batch [0199]/[3759]	Speed: 69.571572 samples/sec	 accuracy=54.570312	 loss=1.898758	 lr=0.010000
Epoch[034] Batch [0249]/[3759]	Speed: 70.891408 samples/sec	 accuracy=54.568750	 loss=1.893556	 lr=0.010000
Epoch[034] Batch [0299]/[3759]	Speed: 70.340881 samples/sec	 accuracy=54.640625	 loss=1.896581	 lr=0.010000
Epoch[034] Batch [0349]/[3759]	Speed: 68.643717 samples/sec	 accuracy=54.540179	 loss=1.906421	 lr=0.010000
Epoch[034] Batch [0399]/[3759]	Speed: 69.696047 samples/sec	 accuracy=54.546875	 loss=1.907192	 lr=0.010000
Epoch[034] Batch [0449]/[3759]	Speed: 70.835671 samples/sec	 accuracy=54.593750	 loss=1.903093	 lr=0.010000
Epoch[034] Batch [0499]/[3759]	Speed: 68.007065 samples/sec	 accuracy=54.440625	 loss=1.905556	 lr=0.010000
Epoch[034] Batch [0549]/[3759]	Speed: 69.664572 samples/sec	 accuracy=54.480114	 loss=1.907813	 lr=0.010000
Epoch[034] Batch [0599]/[3759]	Speed: 70.563000 samples/sec	 accuracy=54.505208	 loss=1.908164	 lr=0.010000
Epoch[034] Batch [0649]/[3759]	Speed: 69.178541 samples/sec	 accuracy=54.540865	 loss=1.907778	 lr=0.010000
Epoch[034] Batch [0699]/[3759]	Speed: 70.355952 samples/sec	 accuracy=54.455357	 loss=1.909044	 lr=0.010000
Epoch[034] Batch [0749]/[3759]	Speed: 69.724276 samples/sec	 accuracy=54.458333	 loss=1.911876	 lr=0.010000
Epoch[034] Batch [0799]/[3759]	Speed: 70.418385 samples/sec	 accuracy=54.464844	 loss=1.914713	 lr=0.010000
Epoch[034] Batch [0849]/[3759]	Speed: 69.737277 samples/sec	 accuracy=54.558824	 loss=1.912756	 lr=0.010000
Epoch[034] Batch [0899]/[3759]	Speed: 69.935678 samples/sec	 accuracy=54.543403	 loss=1.915470	 lr=0.010000
Epoch[034] Batch [0949]/[3759]	Speed: 69.947504 samples/sec	 accuracy=54.544408	 loss=1.916303	 lr=0.010000
Epoch[034] Batch [0999]/[3759]	Speed: 69.167235 samples/sec	 accuracy=54.554688	 loss=1.915571	 lr=0.010000
Epoch[034] Batch [1049]/[3759]	Speed: 70.212467 samples/sec	 accuracy=54.498512	 loss=1.918409	 lr=0.010000
Epoch[034] Batch [1099]/[3759]	Speed: 69.693620 samples/sec	 accuracy=54.522727	 loss=1.915869	 lr=0.010000
Epoch[034] Batch [1149]/[3759]	Speed: 70.322394 samples/sec	 accuracy=54.480978	 loss=1.918417	 lr=0.010000
Epoch[034] Batch [1199]/[3759]	Speed: 68.444469 samples/sec	 accuracy=54.460937	 loss=1.918680	 lr=0.010000
Epoch[034] Batch [1249]/[3759]	Speed: 70.961660 samples/sec	 accuracy=54.447500	 loss=1.917610	 lr=0.010000
Epoch[034] Batch [1299]/[3759]	Speed: 70.485977 samples/sec	 accuracy=54.406250	 loss=1.917858	 lr=0.010000
Epoch[034] Batch [1349]/[3759]	Speed: 69.544094 samples/sec	 accuracy=54.438657	 loss=1.917589	 lr=0.010000
Epoch[034] Batch [1399]/[3759]	Speed: 68.894622 samples/sec	 accuracy=54.447545	 loss=1.917482	 lr=0.010000
Epoch[034] Batch [1449]/[3759]	Speed: 69.457662 samples/sec	 accuracy=54.468750	 loss=1.916763	 lr=0.010000
Epoch[034] Batch [1499]/[3759]	Speed: 70.641604 samples/sec	 accuracy=54.420833	 loss=1.917536	 lr=0.010000
Epoch[034] Batch [1549]/[3759]	Speed: 69.586634 samples/sec	 accuracy=54.459677	 loss=1.917544	 lr=0.010000
Epoch[034] Batch [1599]/[3759]	Speed: 70.577532 samples/sec	 accuracy=54.461914	 loss=1.919011	 lr=0.010000
Epoch[034] Batch [1649]/[3759]	Speed: 68.445078 samples/sec	 accuracy=54.445076	 loss=1.919346	 lr=0.010000
Epoch[034] Batch [1699]/[3759]	Speed: 70.444761 samples/sec	 accuracy=54.464154	 loss=1.919293	 lr=0.010000
Epoch[034] Batch [1749]/[3759]	Speed: 68.399261 samples/sec	 accuracy=54.447321	 loss=1.919487	 lr=0.010000
Epoch[034] Batch [1799]/[3759]	Speed: 71.296375 samples/sec	 accuracy=54.445312	 loss=1.919451	 lr=0.010000
Epoch[034] Batch [1849]/[3759]	Speed: 68.723362 samples/sec	 accuracy=54.392736	 loss=1.919861	 lr=0.010000
Epoch[034] Batch [1899]/[3759]	Speed: 69.984364 samples/sec	 accuracy=54.349507	 loss=1.922343	 lr=0.010000
Epoch[034] Batch [1949]/[3759]	Speed: 70.173947 samples/sec	 accuracy=54.330929	 loss=1.923469	 lr=0.010000
Epoch[034] Batch [1999]/[3759]	Speed: 69.507350 samples/sec	 accuracy=54.303125	 loss=1.924878	 lr=0.010000
Epoch[034] Batch [2049]/[3759]	Speed: 69.967449 samples/sec	 accuracy=54.284299	 loss=1.924854	 lr=0.010000
Epoch[034] Batch [2099]/[3759]	Speed: 70.509566 samples/sec	 accuracy=54.322173	 loss=1.923331	 lr=0.010000
Epoch[034] Batch [2149]/[3759]	Speed: 68.956273 samples/sec	 accuracy=54.300145	 loss=1.924493	 lr=0.010000
Epoch[034] Batch [2199]/[3759]	Speed: 70.558539 samples/sec	 accuracy=54.311790	 loss=1.923861	 lr=0.010000
Epoch[034] Batch [2249]/[3759]	Speed: 68.300456 samples/sec	 accuracy=54.286806	 loss=1.924955	 lr=0.010000
Epoch[034] Batch [2299]/[3759]	Speed: 70.082118 samples/sec	 accuracy=54.279891	 loss=1.925508	 lr=0.010000
Epoch[034] Batch [2349]/[3759]	Speed: 69.239090 samples/sec	 accuracy=54.298537	 loss=1.924864	 lr=0.010000
Epoch[034] Batch [2399]/[3759]	Speed: 71.473528 samples/sec	 accuracy=54.280599	 loss=1.926062	 lr=0.010000
Epoch[034] Batch [2449]/[3759]	Speed: 67.010291 samples/sec	 accuracy=54.290816	 loss=1.924930	 lr=0.010000
Epoch[034] Batch [2499]/[3759]	Speed: 71.170773 samples/sec	 accuracy=54.294375	 loss=1.924695	 lr=0.010000
Epoch[034] Batch [2549]/[3759]	Speed: 70.639680 samples/sec	 accuracy=54.270221	 loss=1.926042	 lr=0.010000
Epoch[034] Batch [2599]/[3759]	Speed: 69.644987 samples/sec	 accuracy=54.262019	 loss=1.926536	 lr=0.010000
Epoch[034] Batch [2649]/[3759]	Speed: 69.419186 samples/sec	 accuracy=54.264151	 loss=1.926941	 lr=0.010000
Epoch[034] Batch [2699]/[3759]	Speed: 67.891658 samples/sec	 accuracy=54.292824	 loss=1.926746	 lr=0.010000
Epoch[034] Batch [2749]/[3759]	Speed: 69.327951 samples/sec	 accuracy=54.290909	 loss=1.926584	 lr=0.010000
Epoch[034] Batch [2799]/[3759]	Speed: 70.892484 samples/sec	 accuracy=54.261719	 loss=1.927694	 lr=0.010000
Epoch[034] Batch [2849]/[3759]	Speed: 69.956097 samples/sec	 accuracy=54.263706	 loss=1.927938	 lr=0.010000
Epoch[034] Batch [2899]/[3759]	Speed: 69.861243 samples/sec	 accuracy=54.255388	 loss=1.928772	 lr=0.010000
Epoch[034] Batch [2949]/[3759]	Speed: 69.442778 samples/sec	 accuracy=54.272246	 loss=1.928244	 lr=0.010000
Epoch[034] Batch [2999]/[3759]	Speed: 70.673941 samples/sec	 accuracy=54.268229	 loss=1.928229	 lr=0.010000
Epoch[034] Batch [3049]/[3759]	Speed: 68.620988 samples/sec	 accuracy=54.229508	 loss=1.930339	 lr=0.010000
Epoch[034] Batch [3099]/[3759]	Speed: 70.299978 samples/sec	 accuracy=54.194052	 loss=1.931629	 lr=0.010000
Epoch[034] Batch [3149]/[3759]	Speed: 70.196947 samples/sec	 accuracy=54.166171	 loss=1.932981	 lr=0.010000
Epoch[034] Batch [3199]/[3759]	Speed: 69.858914 samples/sec	 accuracy=54.160645	 loss=1.933170	 lr=0.010000
Epoch[034] Batch [3249]/[3759]	Speed: 69.406380 samples/sec	 accuracy=54.118750	 loss=1.934714	 lr=0.010000
Epoch[034] Batch [3299]/[3759]	Speed: 68.788374 samples/sec	 accuracy=54.111742	 loss=1.935565	 lr=0.010000
Epoch[034] Batch [3349]/[3759]	Speed: 68.448539 samples/sec	 accuracy=54.106810	 loss=1.935762	 lr=0.010000
Epoch[034] Batch [3399]/[3759]	Speed: 71.054315 samples/sec	 accuracy=54.089154	 loss=1.936256	 lr=0.010000
Epoch[034] Batch [3449]/[3759]	Speed: 70.016877 samples/sec	 accuracy=54.090580	 loss=1.936458	 lr=0.010000
Epoch[034] Batch [3499]/[3759]	Speed: 69.622432 samples/sec	 accuracy=54.093750	 loss=1.936618	 lr=0.010000
Epoch[034] Batch [3549]/[3759]	Speed: 70.374727 samples/sec	 accuracy=54.077025	 loss=1.937262	 lr=0.010000
Epoch[034] Batch [3599]/[3759]	Speed: 69.784020 samples/sec	 accuracy=54.075955	 loss=1.937802	 lr=0.010000
Epoch[034] Batch [3649]/[3759]	Speed: 70.480522 samples/sec	 accuracy=54.057791	 loss=1.938594	 lr=0.010000
Epoch[034] Batch [3699]/[3759]	Speed: 68.545759 samples/sec	 accuracy=54.055321	 loss=1.939185	 lr=0.010000
Epoch[034] Batch [3749]/[3759]	Speed: 78.508907 samples/sec	 accuracy=54.043750	 loss=1.939627	 lr=0.010000
Batch [0049]/[0303]: acc-top1=54.281250 acc-top5=78.281250
Batch [0099]/[0303]: acc-top1=54.250000 acc-top5=78.437500
Batch [0149]/[0303]: acc-top1=55.031250 acc-top5=78.791667
Batch [0199]/[0303]: acc-top1=55.031250 acc-top5=78.937500
Batch [0249]/[0303]: acc-top1=55.231250 acc-top5=79.206250
Batch [0299]/[0303]: acc-top1=54.979167 acc-top5=79.036458
[Epoch 034] training: accuracy=54.048617	 loss=1.939711
[Epoch 034] speed: 69 samples/sec	time cost: 3758.468267
[Epoch 034] validation: acc-top1=54.940182 acc-top5=78.986180 loss=1.981978
Epoch[035] Batch [0049]/[3760]	Speed: 47.412173 samples/sec	 accuracy=54.000000	 loss=1.910897	 lr=0.010000
Epoch[035] Batch [0099]/[3760]	Speed: 68.938389 samples/sec	 accuracy=54.484375	 loss=1.923873	 lr=0.010000
Epoch[035] Batch [0149]/[3760]	Speed: 68.945087 samples/sec	 accuracy=54.947917	 loss=1.900884	 lr=0.010000
Epoch[035] Batch [0199]/[3760]	Speed: 68.348807 samples/sec	 accuracy=55.085938	 loss=1.900487	 lr=0.010000
Epoch[035] Batch [0249]/[3760]	Speed: 71.053217 samples/sec	 accuracy=54.743750	 loss=1.902198	 lr=0.010000
Epoch[035] Batch [0299]/[3760]	Speed: 68.932851 samples/sec	 accuracy=54.994792	 loss=1.891655	 lr=0.010000
Epoch[035] Batch [0349]/[3760]	Speed: 70.031981 samples/sec	 accuracy=54.955357	 loss=1.895759	 lr=0.010000
Epoch[035] Batch [0399]/[3760]	Speed: 69.215684 samples/sec	 accuracy=54.910156	 loss=1.895790	 lr=0.010000
Epoch[035] Batch [0449]/[3760]	Speed: 70.366951 samples/sec	 accuracy=54.843750	 loss=1.903985	 lr=0.010000
Epoch[035] Batch [0499]/[3760]	Speed: 70.195099 samples/sec	 accuracy=54.978125	 loss=1.901773	 lr=0.010000
Epoch[035] Batch [0549]/[3760]	Speed: 69.778217 samples/sec	 accuracy=54.914773	 loss=1.905619	 lr=0.010000
Epoch[035] Batch [0599]/[3760]	Speed: 68.149889 samples/sec	 accuracy=54.984375	 loss=1.905626	 lr=0.010000
Epoch[035] Batch [0649]/[3760]	Speed: 70.491396 samples/sec	 accuracy=54.817308	 loss=1.907950	 lr=0.010000
Epoch[035] Batch [0699]/[3760]	Speed: 71.184706 samples/sec	 accuracy=54.854911	 loss=1.909015	 lr=0.010000
Epoch[035] Batch [0749]/[3760]	Speed: 68.584061 samples/sec	 accuracy=54.902083	 loss=1.904469	 lr=0.010000
Epoch[035] Batch [0799]/[3760]	Speed: 70.316827 samples/sec	 accuracy=54.820312	 loss=1.909375	 lr=0.010000
Epoch[035] Batch [0849]/[3760]	Speed: 69.419855 samples/sec	 accuracy=54.845588	 loss=1.909271	 lr=0.010000
Epoch[035] Batch [0899]/[3760]	Speed: 69.978251 samples/sec	 accuracy=54.737847	 loss=1.912573	 lr=0.010000
Epoch[035] Batch [0949]/[3760]	Speed: 68.487133 samples/sec	 accuracy=54.710526	 loss=1.914466	 lr=0.010000
Epoch[035] Batch [0999]/[3760]	Speed: 70.257806 samples/sec	 accuracy=54.804688	 loss=1.913177	 lr=0.010000
Epoch[035] Batch [1049]/[3760]	Speed: 70.531002 samples/sec	 accuracy=54.794643	 loss=1.915044	 lr=0.010000
Epoch[035] Batch [1099]/[3760]	Speed: 68.859718 samples/sec	 accuracy=54.816761	 loss=1.914884	 lr=0.010000
Epoch[035] Batch [1149]/[3760]	Speed: 70.659049 samples/sec	 accuracy=54.785326	 loss=1.916998	 lr=0.010000
Epoch[035] Batch [1199]/[3760]	Speed: 69.242382 samples/sec	 accuracy=54.760417	 loss=1.918630	 lr=0.010000
Epoch[035] Batch [1249]/[3760]	Speed: 69.846014 samples/sec	 accuracy=54.745000	 loss=1.918757	 lr=0.010000
Epoch[035] Batch [1299]/[3760]	Speed: 69.099712 samples/sec	 accuracy=54.697115	 loss=1.920173	 lr=0.010000
Epoch[035] Batch [1349]/[3760]	Speed: 71.345010 samples/sec	 accuracy=54.657407	 loss=1.921982	 lr=0.010000
Epoch[035] Batch [1399]/[3760]	Speed: 70.408543 samples/sec	 accuracy=54.656250	 loss=1.923935	 lr=0.010000
Epoch[035] Batch [1449]/[3760]	Speed: 67.022129 samples/sec	 accuracy=54.673491	 loss=1.922872	 lr=0.010000
Epoch[035] Batch [1499]/[3760]	Speed: 69.992327 samples/sec	 accuracy=54.705208	 loss=1.923414	 lr=0.010000
Epoch[035] Batch [1549]/[3760]	Speed: 70.283026 samples/sec	 accuracy=54.641129	 loss=1.925459	 lr=0.010000
Epoch[035] Batch [1599]/[3760]	Speed: 69.726965 samples/sec	 accuracy=54.624023	 loss=1.925949	 lr=0.010000
Epoch[035] Batch [1649]/[3760]	Speed: 70.436180 samples/sec	 accuracy=54.600379	 loss=1.925212	 lr=0.010000
Epoch[035] Batch [1699]/[3760]	Speed: 68.709829 samples/sec	 accuracy=54.562500	 loss=1.925907	 lr=0.010000
Epoch[035] Batch [1749]/[3760]	Speed: 70.603474 samples/sec	 accuracy=54.530357	 loss=1.927037	 lr=0.010000
Epoch[035] Batch [1799]/[3760]	Speed: 70.482203 samples/sec	 accuracy=54.499132	 loss=1.927862	 lr=0.010000
Epoch[035] Batch [1849]/[3760]	Speed: 68.810550 samples/sec	 accuracy=54.451858	 loss=1.929694	 lr=0.010000
Epoch[035] Batch [1899]/[3760]	Speed: 67.968532 samples/sec	 accuracy=54.466283	 loss=1.929921	 lr=0.010000
Epoch[035] Batch [1949]/[3760]	Speed: 70.586840 samples/sec	 accuracy=54.443910	 loss=1.931071	 lr=0.010000
Epoch[035] Batch [1999]/[3760]	Speed: 70.099749 samples/sec	 accuracy=54.450781	 loss=1.931739	 lr=0.010000
Epoch[035] Batch [2049]/[3760]	Speed: 68.990029 samples/sec	 accuracy=54.467226	 loss=1.930462	 lr=0.010000
Epoch[035] Batch [2099]/[3760]	Speed: 69.906465 samples/sec	 accuracy=54.444196	 loss=1.931685	 lr=0.010000
Epoch[035] Batch [2149]/[3760]	Speed: 69.470489 samples/sec	 accuracy=54.450581	 loss=1.931256	 lr=0.010000
Epoch[035] Batch [2199]/[3760]	Speed: 69.150646 samples/sec	 accuracy=54.449574	 loss=1.930084	 lr=0.010000
Epoch[035] Batch [2249]/[3760]	Speed: 68.869550 samples/sec	 accuracy=54.419444	 loss=1.930902	 lr=0.010000
Epoch[035] Batch [2299]/[3760]	Speed: 70.337405 samples/sec	 accuracy=54.438859	 loss=1.930720	 lr=0.010000
Epoch[035] Batch [2349]/[3760]	Speed: 69.227736 samples/sec	 accuracy=54.451463	 loss=1.930549	 lr=0.010000
Epoch[035] Batch [2399]/[3760]	Speed: 69.006706 samples/sec	 accuracy=54.432292	 loss=1.931507	 lr=0.010000
Epoch[035] Batch [2449]/[3760]	Speed: 68.813910 samples/sec	 accuracy=54.422194	 loss=1.931404	 lr=0.010000
Epoch[035] Batch [2499]/[3760]	Speed: 70.919822 samples/sec	 accuracy=54.413750	 loss=1.932190	 lr=0.010000
Epoch[035] Batch [2549]/[3760]	Speed: 66.234399 samples/sec	 accuracy=54.398284	 loss=1.932222	 lr=0.010000
Epoch[035] Batch [2599]/[3760]	Speed: 70.813728 samples/sec	 accuracy=54.374399	 loss=1.933545	 lr=0.010000
Epoch[035] Batch [2649]/[3760]	Speed: 70.190830 samples/sec	 accuracy=54.362028	 loss=1.933345	 lr=0.010000
Epoch[035] Batch [2699]/[3760]	Speed: 67.814308 samples/sec	 accuracy=54.351852	 loss=1.933552	 lr=0.010000
Epoch[035] Batch [2749]/[3760]	Speed: 71.011180 samples/sec	 accuracy=54.336932	 loss=1.934282	 lr=0.010000
Epoch[035] Batch [2799]/[3760]	Speed: 70.547281 samples/sec	 accuracy=54.336496	 loss=1.934250	 lr=0.010000
Epoch[035] Batch [2849]/[3760]	Speed: 68.557098 samples/sec	 accuracy=54.322917	 loss=1.934841	 lr=0.010000
Epoch[035] Batch [2899]/[3760]	Speed: 71.343814 samples/sec	 accuracy=54.345366	 loss=1.934108	 lr=0.010000
Epoch[035] Batch [2949]/[3760]	Speed: 67.515204 samples/sec	 accuracy=54.322564	 loss=1.934534	 lr=0.010000
Epoch[035] Batch [2999]/[3760]	Speed: 70.846752 samples/sec	 accuracy=54.310417	 loss=1.934850	 lr=0.010000
Epoch[035] Batch [3049]/[3760]	Speed: 69.417708 samples/sec	 accuracy=54.286885	 loss=1.935067	 lr=0.010000
Epoch[035] Batch [3099]/[3760]	Speed: 69.550988 samples/sec	 accuracy=54.285786	 loss=1.935087	 lr=0.010000
Epoch[035] Batch [3149]/[3760]	Speed: 70.319519 samples/sec	 accuracy=54.283730	 loss=1.935290	 lr=0.010000
Epoch[035] Batch [3199]/[3760]	Speed: 69.870983 samples/sec	 accuracy=54.282715	 loss=1.935185	 lr=0.010000
Epoch[035] Batch [3249]/[3760]	Speed: 69.048271 samples/sec	 accuracy=54.306731	 loss=1.934123	 lr=0.010000
Epoch[035] Batch [3299]/[3760]	Speed: 68.107940 samples/sec	 accuracy=54.294508	 loss=1.934743	 lr=0.010000
Epoch[035] Batch [3349]/[3760]	Speed: 69.790705 samples/sec	 accuracy=54.301306	 loss=1.935034	 lr=0.010000
Epoch[035] Batch [3399]/[3760]	Speed: 70.650153 samples/sec	 accuracy=54.255974	 loss=1.935950	 lr=0.010000
Epoch[035] Batch [3449]/[3760]	Speed: 68.629385 samples/sec	 accuracy=54.240036	 loss=1.936458	 lr=0.010000
Epoch[035] Batch [3499]/[3760]	Speed: 68.814231 samples/sec	 accuracy=54.244196	 loss=1.936325	 lr=0.010000
Epoch[035] Batch [3549]/[3760]	Speed: 69.220284 samples/sec	 accuracy=54.257042	 loss=1.935943	 lr=0.010000
Epoch[035] Batch [3599]/[3760]	Speed: 68.631796 samples/sec	 accuracy=54.251302	 loss=1.936339	 lr=0.010000
Epoch[035] Batch [3649]/[3760]	Speed: 69.481849 samples/sec	 accuracy=54.235445	 loss=1.936329	 lr=0.010000
Epoch[035] Batch [3699]/[3760]	Speed: 68.857549 samples/sec	 accuracy=54.238176	 loss=1.936041	 lr=0.010000
Epoch[035] Batch [3749]/[3760]	Speed: 78.257519 samples/sec	 accuracy=54.244167	 loss=1.935880	 lr=0.010000
Batch [0049]/[0303]: acc-top1=52.718750 acc-top5=78.281250
Batch [0099]/[0303]: acc-top1=52.937500 acc-top5=78.187500
Batch [0149]/[0303]: acc-top1=53.447917 acc-top5=78.291667
Batch [0199]/[0303]: acc-top1=53.867188 acc-top5=78.507812
Batch [0249]/[0303]: acc-top1=53.993750 acc-top5=78.675000
Batch [0299]/[0303]: acc-top1=53.932292 acc-top5=78.380208
[Epoch 035] training: accuracy=54.256981	 loss=1.935538
[Epoch 035] speed: 69 samples/sec	time cost: 3765.418977
[Epoch 035] validation: acc-top1=53.888201 acc-top5=78.357054 loss=2.045827
Epoch[036] Batch [0049]/[3760]	Speed: 49.359299 samples/sec	 accuracy=55.750000	 loss=1.839590	 lr=0.010000
Epoch[036] Batch [0099]/[3760]	Speed: 67.672244 samples/sec	 accuracy=54.843750	 loss=1.882685	 lr=0.010000
Epoch[036] Batch [0149]/[3760]	Speed: 66.704609 samples/sec	 accuracy=55.020833	 loss=1.879656	 lr=0.010000
Epoch[036] Batch [0199]/[3760]	Speed: 72.283188 samples/sec	 accuracy=54.789062	 loss=1.894684	 lr=0.010000
Epoch[036] Batch [0249]/[3760]	Speed: 68.892615 samples/sec	 accuracy=54.718750	 loss=1.912899	 lr=0.010000
Epoch[036] Batch [0299]/[3760]	Speed: 69.457496 samples/sec	 accuracy=54.640625	 loss=1.908834	 lr=0.010000
Epoch[036] Batch [0349]/[3760]	Speed: 70.045116 samples/sec	 accuracy=54.544643	 loss=1.910766	 lr=0.010000
Epoch[036] Batch [0399]/[3760]	Speed: 70.633390 samples/sec	 accuracy=54.554688	 loss=1.909438	 lr=0.010000
Epoch[036] Batch [0449]/[3760]	Speed: 69.123524 samples/sec	 accuracy=54.475694	 loss=1.915556	 lr=0.010000
Epoch[036] Batch [0499]/[3760]	Speed: 68.938314 samples/sec	 accuracy=54.265625	 loss=1.923295	 lr=0.010000
Epoch[036] Batch [0549]/[3760]	Speed: 69.912350 samples/sec	 accuracy=54.289773	 loss=1.920563	 lr=0.010000
Epoch[036] Batch [0599]/[3760]	Speed: 69.382495 samples/sec	 accuracy=54.213542	 loss=1.925062	 lr=0.010000
Epoch[036] Batch [0649]/[3760]	Speed: 68.080272 samples/sec	 accuracy=54.233173	 loss=1.924072	 lr=0.010000
Epoch[036] Batch [0699]/[3760]	Speed: 70.912121 samples/sec	 accuracy=54.178571	 loss=1.924814	 lr=0.010000
Epoch[036] Batch [0749]/[3760]	Speed: 69.398079 samples/sec	 accuracy=54.233333	 loss=1.922523	 lr=0.010000
Epoch[036] Batch [0799]/[3760]	Speed: 70.493301 samples/sec	 accuracy=54.337891	 loss=1.919928	 lr=0.010000
Epoch[036] Batch [0849]/[3760]	Speed: 69.680324 samples/sec	 accuracy=54.347426	 loss=1.921209	 lr=0.010000
Epoch[036] Batch [0899]/[3760]	Speed: 70.226234 samples/sec	 accuracy=54.282986	 loss=1.924037	 lr=0.010000
Epoch[036] Batch [0949]/[3760]	Speed: 69.264026 samples/sec	 accuracy=54.289474	 loss=1.924131	 lr=0.010000
Epoch[036] Batch [0999]/[3760]	Speed: 69.363601 samples/sec	 accuracy=54.335938	 loss=1.923042	 lr=0.010000
Epoch[036] Batch [1049]/[3760]	Speed: 70.169361 samples/sec	 accuracy=54.311012	 loss=1.922931	 lr=0.010000
Epoch[036] Batch [1099]/[3760]	Speed: 70.170011 samples/sec	 accuracy=54.321023	 loss=1.921743	 lr=0.010000
Epoch[036] Batch [1149]/[3760]	Speed: 69.212537 samples/sec	 accuracy=54.365489	 loss=1.920985	 lr=0.010000
Epoch[036] Batch [1199]/[3760]	Speed: 68.694134 samples/sec	 accuracy=54.376302	 loss=1.921311	 lr=0.010000
Epoch[036] Batch [1249]/[3760]	Speed: 68.581673 samples/sec	 accuracy=54.302500	 loss=1.924927	 lr=0.010000
Epoch[036] Batch [1299]/[3760]	Speed: 68.444908 samples/sec	 accuracy=54.281250	 loss=1.926590	 lr=0.010000
Epoch[036] Batch [1349]/[3760]	Speed: 71.284339 samples/sec	 accuracy=54.296296	 loss=1.925898	 lr=0.010000
Epoch[036] Batch [1399]/[3760]	Speed: 67.999827 samples/sec	 accuracy=54.286830	 loss=1.926159	 lr=0.010000
Epoch[036] Batch [1449]/[3760]	Speed: 69.593982 samples/sec	 accuracy=54.230603	 loss=1.929512	 lr=0.010000
Epoch[036] Batch [1499]/[3760]	Speed: 69.929977 samples/sec	 accuracy=54.223958	 loss=1.929744	 lr=0.010000
Epoch[036] Batch [1549]/[3760]	Speed: 70.132506 samples/sec	 accuracy=54.182460	 loss=1.931683	 lr=0.010000
Epoch[036] Batch [1599]/[3760]	Speed: 71.462583 samples/sec	 accuracy=54.173828	 loss=1.931419	 lr=0.010000
Epoch[036] Batch [1649]/[3760]	Speed: 66.843787 samples/sec	 accuracy=54.128788	 loss=1.933673	 lr=0.010000
Epoch[036] Batch [1699]/[3760]	Speed: 71.704897 samples/sec	 accuracy=54.150735	 loss=1.933636	 lr=0.010000
Epoch[036] Batch [1749]/[3760]	Speed: 70.783240 samples/sec	 accuracy=54.106250	 loss=1.935909	 lr=0.010000
Epoch[036] Batch [1799]/[3760]	Speed: 69.368421 samples/sec	 accuracy=54.078993	 loss=1.936744	 lr=0.010000
Epoch[036] Batch [1849]/[3760]	Speed: 69.549838 samples/sec	 accuracy=54.075169	 loss=1.937348	 lr=0.010000
Epoch[036] Batch [1899]/[3760]	Speed: 67.356373 samples/sec	 accuracy=54.117599	 loss=1.936178	 lr=0.010000
Epoch[036] Batch [1949]/[3760]	Speed: 68.788712 samples/sec	 accuracy=54.107372	 loss=1.936285	 lr=0.010000
Epoch[036] Batch [1999]/[3760]	Speed: 70.639961 samples/sec	 accuracy=54.121875	 loss=1.936208	 lr=0.010000
Epoch[036] Batch [2049]/[3760]	Speed: 69.172742 samples/sec	 accuracy=54.142530	 loss=1.935441	 lr=0.010000
Epoch[036] Batch [2099]/[3760]	Speed: 70.233879 samples/sec	 accuracy=54.098958	 loss=1.937243	 lr=0.010000
Epoch[036] Batch [2149]/[3760]	Speed: 70.514690 samples/sec	 accuracy=54.111919	 loss=1.937131	 lr=0.010000
Epoch[036] Batch [2199]/[3760]	Speed: 68.556606 samples/sec	 accuracy=54.107955	 loss=1.937325	 lr=0.010000
Epoch[036] Batch [2249]/[3760]	Speed: 69.868859 samples/sec	 accuracy=54.097222	 loss=1.938879	 lr=0.010000
Epoch[036] Batch [2299]/[3760]	Speed: 70.356568 samples/sec	 accuracy=54.108016	 loss=1.937933	 lr=0.010000
Epoch[036] Batch [2349]/[3760]	Speed: 68.254357 samples/sec	 accuracy=54.117021	 loss=1.938469	 lr=0.010000
Epoch[036] Batch [2399]/[3760]	Speed: 69.680528 samples/sec	 accuracy=54.120443	 loss=1.938481	 lr=0.010000
Epoch[036] Batch [2449]/[3760]	Speed: 69.764815 samples/sec	 accuracy=54.097577	 loss=1.938396	 lr=0.010000
Epoch[036] Batch [2499]/[3760]	Speed: 69.840399 samples/sec	 accuracy=54.071875	 loss=1.939320	 lr=0.010000
Epoch[036] Batch [2549]/[3760]	Speed: 70.243501 samples/sec	 accuracy=54.075368	 loss=1.938257	 lr=0.010000
Epoch[036] Batch [2599]/[3760]	Speed: 70.451046 samples/sec	 accuracy=54.067909	 loss=1.939023	 lr=0.010000
Epoch[036] Batch [2649]/[3760]	Speed: 69.145291 samples/sec	 accuracy=54.067217	 loss=1.939423	 lr=0.010000
Epoch[036] Batch [2699]/[3760]	Speed: 69.340768 samples/sec	 accuracy=54.057292	 loss=1.939830	 lr=0.010000
Epoch[036] Batch [2749]/[3760]	Speed: 68.320537 samples/sec	 accuracy=54.040341	 loss=1.941786	 lr=0.010000
Epoch[036] Batch [2799]/[3760]	Speed: 70.318872 samples/sec	 accuracy=54.049665	 loss=1.941270	 lr=0.010000
Epoch[036] Batch [2849]/[3760]	Speed: 69.995131 samples/sec	 accuracy=54.059759	 loss=1.941145	 lr=0.010000
Epoch[036] Batch [2899]/[3760]	Speed: 68.752443 samples/sec	 accuracy=54.072737	 loss=1.940776	 lr=0.010000
Epoch[036] Batch [2949]/[3760]	Speed: 70.225073 samples/sec	 accuracy=54.050318	 loss=1.941137	 lr=0.010000
Epoch[036] Batch [2999]/[3760]	Speed: 69.624131 samples/sec	 accuracy=54.048958	 loss=1.941545	 lr=0.010000
Epoch[036] Batch [3049]/[3760]	Speed: 69.690433 samples/sec	 accuracy=54.064549	 loss=1.941459	 lr=0.010000
Epoch[036] Batch [3099]/[3760]	Speed: 70.474307 samples/sec	 accuracy=54.103327	 loss=1.940579	 lr=0.010000
Epoch[036] Batch [3149]/[3760]	Speed: 69.123727 samples/sec	 accuracy=54.097718	 loss=1.940601	 lr=0.010000
Epoch[036] Batch [3199]/[3760]	Speed: 68.431532 samples/sec	 accuracy=54.131836	 loss=1.939909	 lr=0.010000
Epoch[036] Batch [3249]/[3760]	Speed: 68.904532 samples/sec	 accuracy=54.103365	 loss=1.940592	 lr=0.010000
Epoch[036] Batch [3299]/[3760]	Speed: 70.763705 samples/sec	 accuracy=54.116477	 loss=1.939797	 lr=0.010000
Epoch[036] Batch [3349]/[3760]	Speed: 70.930690 samples/sec	 accuracy=54.099813	 loss=1.940489	 lr=0.010000
Epoch[036] Batch [3399]/[3760]	Speed: 67.423201 samples/sec	 accuracy=54.123621	 loss=1.939350	 lr=0.010000
Epoch[036] Batch [3449]/[3760]	Speed: 71.640077 samples/sec	 accuracy=54.118659	 loss=1.939420	 lr=0.010000
Epoch[036] Batch [3499]/[3760]	Speed: 68.310214 samples/sec	 accuracy=54.120536	 loss=1.939110	 lr=0.010000
Epoch[036] Batch [3549]/[3760]	Speed: 70.902853 samples/sec	 accuracy=54.115757	 loss=1.939069	 lr=0.010000
Epoch[036] Batch [3599]/[3760]	Speed: 70.316697 samples/sec	 accuracy=54.134983	 loss=1.938393	 lr=0.010000
Epoch[036] Batch [3649]/[3760]	Speed: 69.114188 samples/sec	 accuracy=54.148116	 loss=1.938334	 lr=0.010000
Epoch[036] Batch [3699]/[3760]	Speed: 69.513413 samples/sec	 accuracy=54.145693	 loss=1.938886	 lr=0.010000
Epoch[036] Batch [3749]/[3760]	Speed: 76.404753 samples/sec	 accuracy=54.161250	 loss=1.938702	 lr=0.010000
Batch [0049]/[0303]: acc-top1=53.875000 acc-top5=78.281250
Batch [0099]/[0303]: acc-top1=53.828125 acc-top5=78.343750
Batch [0149]/[0303]: acc-top1=54.239583 acc-top5=78.989583
Batch [0199]/[0303]: acc-top1=54.562500 acc-top5=79.078125
Batch [0249]/[0303]: acc-top1=54.825000 acc-top5=79.387500
Batch [0299]/[0303]: acc-top1=54.682292 acc-top5=79.135417
[Epoch 036] training: accuracy=54.152261	 loss=1.938754
[Epoch 036] speed: 69 samples/sec	time cost: 3764.700441
[Epoch 036] validation: acc-top1=54.672030 acc-top5=79.130569 loss=1.997851
Epoch[037] Batch [0049]/[3759]	Speed: 48.253808 samples/sec	 accuracy=53.656250	 loss=1.932206	 lr=0.010000
Epoch[037] Batch [0099]/[3759]	Speed: 68.307174 samples/sec	 accuracy=54.859375	 loss=1.899286	 lr=0.010000
Epoch[037] Batch [0149]/[3759]	Speed: 68.135969 samples/sec	 accuracy=54.906250	 loss=1.910978	 lr=0.010000
Epoch[037] Batch [0199]/[3759]	Speed: 70.479044 samples/sec	 accuracy=54.984375	 loss=1.908678	 lr=0.010000
Epoch[037] Batch [0249]/[3759]	Speed: 68.199750 samples/sec	 accuracy=54.831250	 loss=1.913303	 lr=0.010000
Epoch[037] Batch [0299]/[3759]	Speed: 71.173462 samples/sec	 accuracy=54.661458	 loss=1.921481	 lr=0.010000
Epoch[037] Batch [0349]/[3759]	Speed: 69.869248 samples/sec	 accuracy=54.566964	 loss=1.923925	 lr=0.010000
Epoch[037] Batch [0399]/[3759]	Speed: 69.323193 samples/sec	 accuracy=54.734375	 loss=1.916591	 lr=0.010000
Epoch[037] Batch [0449]/[3759]	Speed: 70.597101 samples/sec	 accuracy=54.819444	 loss=1.914081	 lr=0.010000
Epoch[037] Batch [0499]/[3759]	Speed: 68.717832 samples/sec	 accuracy=54.759375	 loss=1.913533	 lr=0.010000
Epoch[037] Batch [0549]/[3759]	Speed: 68.808676 samples/sec	 accuracy=54.767045	 loss=1.910603	 lr=0.010000
Epoch[037] Batch [0599]/[3759]	Speed: 69.237832 samples/sec	 accuracy=54.861979	 loss=1.906249	 lr=0.010000
Epoch[037] Batch [0649]/[3759]	Speed: 71.118029 samples/sec	 accuracy=54.841346	 loss=1.906224	 lr=0.010000
Epoch[037] Batch [0699]/[3759]	Speed: 67.947785 samples/sec	 accuracy=54.919643	 loss=1.904845	 lr=0.010000
Epoch[037] Batch [0749]/[3759]	Speed: 70.589184 samples/sec	 accuracy=54.995833	 loss=1.899185	 lr=0.010000
Epoch[037] Batch [0799]/[3759]	Speed: 70.072648 samples/sec	 accuracy=54.984375	 loss=1.900324	 lr=0.010000
Epoch[037] Batch [0849]/[3759]	Speed: 67.956108 samples/sec	 accuracy=54.909926	 loss=1.900816	 lr=0.010000
Epoch[037] Batch [0899]/[3759]	Speed: 71.075523 samples/sec	 accuracy=54.883681	 loss=1.903276	 lr=0.010000
Epoch[037] Batch [0949]/[3759]	Speed: 67.530364 samples/sec	 accuracy=54.927632	 loss=1.903324	 lr=0.010000
Epoch[037] Batch [0999]/[3759]	Speed: 70.123933 samples/sec	 accuracy=54.984375	 loss=1.900579	 lr=0.010000
Epoch[037] Batch [1049]/[3759]	Speed: 70.131189 samples/sec	 accuracy=54.943452	 loss=1.901450	 lr=0.010000
Epoch[037] Batch [1099]/[3759]	Speed: 69.840765 samples/sec	 accuracy=54.894886	 loss=1.901858	 lr=0.010000
Epoch[037] Batch [1149]/[3759]	Speed: 70.217304 samples/sec	 accuracy=54.816576	 loss=1.905441	 lr=0.010000
Epoch[037] Batch [1199]/[3759]	Speed: 68.255290 samples/sec	 accuracy=54.815104	 loss=1.905498	 lr=0.010000
Epoch[037] Batch [1249]/[3759]	Speed: 70.285786 samples/sec	 accuracy=54.872500	 loss=1.904375	 lr=0.010000
Epoch[037] Batch [1299]/[3759]	Speed: 70.903476 samples/sec	 accuracy=54.859375	 loss=1.903765	 lr=0.010000
Epoch[037] Batch [1349]/[3759]	Speed: 69.855683 samples/sec	 accuracy=54.821759	 loss=1.904578	 lr=0.010000
Epoch[037] Batch [1399]/[3759]	Speed: 69.663885 samples/sec	 accuracy=54.829241	 loss=1.903806	 lr=0.010000
Epoch[037] Batch [1449]/[3759]	Speed: 69.364470 samples/sec	 accuracy=54.848060	 loss=1.904401	 lr=0.010000
Epoch[037] Batch [1499]/[3759]	Speed: 71.110039 samples/sec	 accuracy=54.854167	 loss=1.903605	 lr=0.010000
Epoch[037] Batch [1549]/[3759]	Speed: 68.315715 samples/sec	 accuracy=54.881048	 loss=1.902650	 lr=0.010000
Epoch[037] Batch [1599]/[3759]	Speed: 69.801796 samples/sec	 accuracy=54.885742	 loss=1.902945	 lr=0.010000
Epoch[037] Batch [1649]/[3759]	Speed: 68.886028 samples/sec	 accuracy=54.838068	 loss=1.904157	 lr=0.010000
Epoch[037] Batch [1699]/[3759]	Speed: 70.126187 samples/sec	 accuracy=54.807904	 loss=1.904814	 lr=0.010000
Epoch[037] Batch [1749]/[3759]	Speed: 70.578134 samples/sec	 accuracy=54.840179	 loss=1.903938	 lr=0.010000
Epoch[037] Batch [1799]/[3759]	Speed: 69.767357 samples/sec	 accuracy=54.855035	 loss=1.904046	 lr=0.010000
Epoch[037] Batch [1849]/[3759]	Speed: 69.482083 samples/sec	 accuracy=54.840372	 loss=1.904878	 lr=0.010000
Epoch[037] Batch [1899]/[3759]	Speed: 68.645260 samples/sec	 accuracy=54.862664	 loss=1.904728	 lr=0.010000
Epoch[037] Batch [1949]/[3759]	Speed: 71.392126 samples/sec	 accuracy=54.854167	 loss=1.905100	 lr=0.010000
Epoch[037] Batch [1999]/[3759]	Speed: 68.296357 samples/sec	 accuracy=54.783594	 loss=1.906766	 lr=0.010000
Epoch[037] Batch [2049]/[3759]	Speed: 70.200860 samples/sec	 accuracy=54.763720	 loss=1.907787	 lr=0.010000
Epoch[037] Batch [2099]/[3759]	Speed: 68.950448 samples/sec	 accuracy=54.758929	 loss=1.908466	 lr=0.010000
Epoch[037] Batch [2149]/[3759]	Speed: 70.455754 samples/sec	 accuracy=54.700581	 loss=1.910324	 lr=0.010000
Epoch[037] Batch [2199]/[3759]	Speed: 69.856059 samples/sec	 accuracy=54.718040	 loss=1.909437	 lr=0.010000
Epoch[037] Batch [2249]/[3759]	Speed: 69.186127 samples/sec	 accuracy=54.728472	 loss=1.909209	 lr=0.010000
Epoch[037] Batch [2299]/[3759]	Speed: 70.102320 samples/sec	 accuracy=54.701766	 loss=1.910310	 lr=0.010000
Epoch[037] Batch [2349]/[3759]	Speed: 70.214603 samples/sec	 accuracy=54.681516	 loss=1.910772	 lr=0.010000
Epoch[037] Batch [2399]/[3759]	Speed: 69.213102 samples/sec	 accuracy=54.665365	 loss=1.910723	 lr=0.010000
Epoch[037] Batch [2449]/[3759]	Speed: 69.826819 samples/sec	 accuracy=54.619898	 loss=1.912362	 lr=0.010000
Epoch[037] Batch [2499]/[3759]	Speed: 69.426471 samples/sec	 accuracy=54.620000	 loss=1.911876	 lr=0.010000
Epoch[037] Batch [2549]/[3759]	Speed: 68.862944 samples/sec	 accuracy=54.643995	 loss=1.911936	 lr=0.010000
Epoch[037] Batch [2599]/[3759]	Speed: 69.548791 samples/sec	 accuracy=54.642428	 loss=1.911671	 lr=0.010000
Epoch[037] Batch [2649]/[3759]	Speed: 71.157761 samples/sec	 accuracy=54.628538	 loss=1.912360	 lr=0.010000
Epoch[037] Batch [2699]/[3759]	Speed: 69.970580 samples/sec	 accuracy=54.641204	 loss=1.911798	 lr=0.010000
Epoch[037] Batch [2749]/[3759]	Speed: 69.250306 samples/sec	 accuracy=54.651705	 loss=1.912138	 lr=0.010000
Epoch[037] Batch [2799]/[3759]	Speed: 68.639099 samples/sec	 accuracy=54.626674	 loss=1.913737	 lr=0.010000
Epoch[037] Batch [2849]/[3759]	Speed: 70.708832 samples/sec	 accuracy=54.602522	 loss=1.914703	 lr=0.010000
Epoch[037] Batch [2899]/[3759]	Speed: 70.345004 samples/sec	 accuracy=54.589978	 loss=1.915409	 lr=0.010000
Epoch[037] Batch [2949]/[3759]	Speed: 69.492243 samples/sec	 accuracy=54.582097	 loss=1.915368	 lr=0.010000
Epoch[037] Batch [2999]/[3759]	Speed: 68.920479 samples/sec	 accuracy=54.570312	 loss=1.915205	 lr=0.010000
Epoch[037] Batch [3049]/[3759]	Speed: 70.020521 samples/sec	 accuracy=54.572746	 loss=1.915042	 lr=0.010000
Epoch[037] Batch [3099]/[3759]	Speed: 70.172089 samples/sec	 accuracy=54.560484	 loss=1.916057	 lr=0.010000
Epoch[037] Batch [3149]/[3759]	Speed: 69.390325 samples/sec	 accuracy=54.550099	 loss=1.916517	 lr=0.010000
Epoch[037] Batch [3199]/[3759]	Speed: 70.266373 samples/sec	 accuracy=54.522461	 loss=1.917298	 lr=0.010000
Epoch[037] Batch [3249]/[3759]	Speed: 70.296181 samples/sec	 accuracy=54.502885	 loss=1.918348	 lr=0.010000
Epoch[037] Batch [3299]/[3759]	Speed: 68.864844 samples/sec	 accuracy=54.516572	 loss=1.918477	 lr=0.010000
Epoch[037] Batch [3349]/[3759]	Speed: 69.906497 samples/sec	 accuracy=54.513060	 loss=1.918873	 lr=0.010000
Epoch[037] Batch [3399]/[3759]	Speed: 70.301141 samples/sec	 accuracy=54.508272	 loss=1.919497	 lr=0.010000
Epoch[037] Batch [3449]/[3759]	Speed: 67.440231 samples/sec	 accuracy=54.519928	 loss=1.919349	 lr=0.010000
Epoch[037] Batch [3499]/[3759]	Speed: 71.724134 samples/sec	 accuracy=54.492411	 loss=1.920742	 lr=0.010000
Epoch[037] Batch [3549]/[3759]	Speed: 70.169011 samples/sec	 accuracy=54.478873	 loss=1.921464	 lr=0.010000
Epoch[037] Batch [3599]/[3759]	Speed: 69.959165 samples/sec	 accuracy=54.470486	 loss=1.922311	 lr=0.010000
Epoch[037] Batch [3649]/[3759]	Speed: 69.838631 samples/sec	 accuracy=54.464897	 loss=1.922735	 lr=0.010000
Epoch[037] Batch [3699]/[3759]	Speed: 68.154675 samples/sec	 accuracy=54.471284	 loss=1.922804	 lr=0.010000
Epoch[037] Batch [3749]/[3759]	Speed: 77.864846 samples/sec	 accuracy=54.473750	 loss=1.922929	 lr=0.010000
Batch [0049]/[0303]: acc-top1=53.250000 acc-top5=78.937500
Batch [0099]/[0303]: acc-top1=53.656250 acc-top5=78.859375
Batch [0149]/[0303]: acc-top1=54.385417 acc-top5=79.062500
Batch [0199]/[0303]: acc-top1=54.898438 acc-top5=79.203125
Batch [0249]/[0303]: acc-top1=55.056250 acc-top5=79.487500
Batch [0299]/[0303]: acc-top1=54.635417 acc-top5=79.203125
[Epoch 037] training: accuracy=54.477587	 loss=1.922707
[Epoch 037] speed: 69 samples/sec	time cost: 3757.324147
[Epoch 037] validation: acc-top1=54.641089 acc-top5=79.176980 loss=2.008608
Epoch[038] Batch [0049]/[3760]	Speed: 47.900882 samples/sec	 accuracy=55.500000	 loss=1.893727	 lr=0.010000
Epoch[038] Batch [0099]/[3760]	Speed: 68.645432 samples/sec	 accuracy=55.812500	 loss=1.865385	 lr=0.010000
Epoch[038] Batch [0149]/[3760]	Speed: 69.428213 samples/sec	 accuracy=55.552083	 loss=1.870666	 lr=0.010000
Epoch[038] Batch [0199]/[3760]	Speed: 70.002962 samples/sec	 accuracy=55.523437	 loss=1.874207	 lr=0.010000
Epoch[038] Batch [0249]/[3760]	Speed: 70.438839 samples/sec	 accuracy=55.200000	 loss=1.885996	 lr=0.010000
Epoch[038] Batch [0299]/[3760]	Speed: 67.668040 samples/sec	 accuracy=55.098958	 loss=1.891754	 lr=0.010000
Epoch[038] Batch [0349]/[3760]	Speed: 69.721712 samples/sec	 accuracy=55.093750	 loss=1.884559	 lr=0.010000
Epoch[038] Batch [0399]/[3760]	Speed: 70.444853 samples/sec	 accuracy=54.980469	 loss=1.892535	 lr=0.010000
Epoch[038] Batch [0449]/[3760]	Speed: 70.818255 samples/sec	 accuracy=54.979167	 loss=1.892652	 lr=0.010000
Epoch[038] Batch [0499]/[3760]	Speed: 68.843961 samples/sec	 accuracy=55.034375	 loss=1.892177	 lr=0.010000
Epoch[038] Batch [0549]/[3760]	Speed: 70.819637 samples/sec	 accuracy=54.914773	 loss=1.892193	 lr=0.010000
Epoch[038] Batch [0599]/[3760]	Speed: 68.282916 samples/sec	 accuracy=54.851562	 loss=1.893697	 lr=0.010000
Epoch[038] Batch [0649]/[3760]	Speed: 70.818659 samples/sec	 accuracy=54.879808	 loss=1.893299	 lr=0.010000
Epoch[038] Batch [0699]/[3760]	Speed: 71.209770 samples/sec	 accuracy=54.875000	 loss=1.894095	 lr=0.010000
Epoch[038] Batch [0749]/[3760]	Speed: 69.946576 samples/sec	 accuracy=54.862500	 loss=1.896338	 lr=0.010000
Epoch[038] Batch [0799]/[3760]	Speed: 68.074908 samples/sec	 accuracy=54.896484	 loss=1.895847	 lr=0.010000
Epoch[038] Batch [0849]/[3760]	Speed: 70.938491 samples/sec	 accuracy=54.871324	 loss=1.896463	 lr=0.010000
Epoch[038] Batch [0899]/[3760]	Speed: 69.287378 samples/sec	 accuracy=54.810764	 loss=1.896559	 lr=0.010000
Epoch[038] Batch [0949]/[3760]	Speed: 69.034240 samples/sec	 accuracy=54.705592	 loss=1.899797	 lr=0.010000
Epoch[038] Batch [0999]/[3760]	Speed: 69.897682 samples/sec	 accuracy=54.690625	 loss=1.897949	 lr=0.010000
Epoch[038] Batch [1049]/[3760]	Speed: 69.811797 samples/sec	 accuracy=54.712798	 loss=1.898818	 lr=0.010000
Epoch[038] Batch [1099]/[3760]	Speed: 70.316134 samples/sec	 accuracy=54.708807	 loss=1.899081	 lr=0.010000
Epoch[038] Batch [1149]/[3760]	Speed: 70.140045 samples/sec	 accuracy=54.646739	 loss=1.902292	 lr=0.010000
Epoch[038] Batch [1199]/[3760]	Speed: 69.483063 samples/sec	 accuracy=54.606771	 loss=1.904143	 lr=0.010000
Epoch[038] Batch [1249]/[3760]	Speed: 70.112345 samples/sec	 accuracy=54.540000	 loss=1.906743	 lr=0.010000
Epoch[038] Batch [1299]/[3760]	Speed: 69.773269 samples/sec	 accuracy=54.507212	 loss=1.909457	 lr=0.010000
Epoch[038] Batch [1349]/[3760]	Speed: 68.247752 samples/sec	 accuracy=54.491898	 loss=1.910516	 lr=0.010000
Epoch[038] Batch [1399]/[3760]	Speed: 70.751343 samples/sec	 accuracy=54.484375	 loss=1.909568	 lr=0.010000
Epoch[038] Batch [1449]/[3760]	Speed: 68.100310 samples/sec	 accuracy=54.492457	 loss=1.910220	 lr=0.010000
Epoch[038] Batch [1499]/[3760]	Speed: 70.801442 samples/sec	 accuracy=54.516667	 loss=1.909591	 lr=0.010000
Epoch[038] Batch [1549]/[3760]	Speed: 70.688576 samples/sec	 accuracy=54.496976	 loss=1.909779	 lr=0.010000
Epoch[038] Batch [1599]/[3760]	Speed: 69.356848 samples/sec	 accuracy=54.506836	 loss=1.910997	 lr=0.010000
Epoch[038] Batch [1649]/[3760]	Speed: 67.802152 samples/sec	 accuracy=54.512311	 loss=1.911736	 lr=0.010000
Epoch[038] Batch [1699]/[3760]	Speed: 70.297474 samples/sec	 accuracy=54.484375	 loss=1.912732	 lr=0.010000
Epoch[038] Batch [1749]/[3760]	Speed: 70.288466 samples/sec	 accuracy=54.525000	 loss=1.912248	 lr=0.010000
Epoch[038] Batch [1799]/[3760]	Speed: 70.236674 samples/sec	 accuracy=54.532118	 loss=1.912633	 lr=0.010000
Epoch[038] Batch [1849]/[3760]	Speed: 69.526745 samples/sec	 accuracy=54.500845	 loss=1.913616	 lr=0.010000
Epoch[038] Batch [1899]/[3760]	Speed: 69.050383 samples/sec	 accuracy=54.491776	 loss=1.914801	 lr=0.010000
Epoch[038] Batch [1949]/[3760]	Speed: 70.382811 samples/sec	 accuracy=54.517628	 loss=1.914401	 lr=0.010000
Epoch[038] Batch [1999]/[3760]	Speed: 68.881082 samples/sec	 accuracy=54.502344	 loss=1.914243	 lr=0.010000
Epoch[038] Batch [2049]/[3760]	Speed: 70.337477 samples/sec	 accuracy=54.528201	 loss=1.913779	 lr=0.010000
Epoch[038] Batch [2099]/[3760]	Speed: 69.711745 samples/sec	 accuracy=54.553571	 loss=1.914329	 lr=0.010000
Epoch[038] Batch [2149]/[3760]	Speed: 68.598803 samples/sec	 accuracy=54.537064	 loss=1.914857	 lr=0.010000
Epoch[038] Batch [2199]/[3760]	Speed: 70.698658 samples/sec	 accuracy=54.516335	 loss=1.915329	 lr=0.010000
Epoch[038] Batch [2249]/[3760]	Speed: 70.155981 samples/sec	 accuracy=54.558333	 loss=1.914495	 lr=0.010000
Epoch[038] Batch [2299]/[3760]	Speed: 69.604768 samples/sec	 accuracy=54.567255	 loss=1.914501	 lr=0.010000
Epoch[038] Batch [2349]/[3760]	Speed: 69.109467 samples/sec	 accuracy=54.565824	 loss=1.914136	 lr=0.010000
Epoch[038] Batch [2399]/[3760]	Speed: 71.179779 samples/sec	 accuracy=54.546224	 loss=1.915843	 lr=0.010000
Epoch[038] Batch [2449]/[3760]	Speed: 69.774696 samples/sec	 accuracy=54.529337	 loss=1.916125	 lr=0.010000
Epoch[038] Batch [2499]/[3760]	Speed: 69.382356 samples/sec	 accuracy=54.530000	 loss=1.915935	 lr=0.010000
Epoch[038] Batch [2549]/[3760]	Speed: 68.148153 samples/sec	 accuracy=54.499387	 loss=1.917158	 lr=0.010000
Epoch[038] Batch [2599]/[3760]	Speed: 69.955242 samples/sec	 accuracy=54.521034	 loss=1.916497	 lr=0.010000
Epoch[038] Batch [2649]/[3760]	Speed: 67.934134 samples/sec	 accuracy=54.497052	 loss=1.916896	 lr=0.010000
Epoch[038] Batch [2699]/[3760]	Speed: 69.985362 samples/sec	 accuracy=54.466435	 loss=1.918600	 lr=0.010000
Epoch[038] Batch [2749]/[3760]	Speed: 69.408861 samples/sec	 accuracy=54.460227	 loss=1.919734	 lr=0.010000
Epoch[038] Batch [2799]/[3760]	Speed: 71.499457 samples/sec	 accuracy=54.453125	 loss=1.919826	 lr=0.010000
Epoch[038] Batch [2849]/[3760]	Speed: 67.247757 samples/sec	 accuracy=54.436404	 loss=1.921379	 lr=0.010000
Epoch[038] Batch [2899]/[3760]	Speed: 71.396903 samples/sec	 accuracy=54.403017	 loss=1.922274	 lr=0.010000
Epoch[038] Batch [2949]/[3760]	Speed: 70.783842 samples/sec	 accuracy=54.427436	 loss=1.920899	 lr=0.010000
Epoch[038] Batch [2999]/[3760]	Speed: 68.768795 samples/sec	 accuracy=54.422917	 loss=1.920733	 lr=0.010000
Epoch[038] Batch [3049]/[3760]	Speed: 69.907896 samples/sec	 accuracy=54.420594	 loss=1.920557	 lr=0.010000
Epoch[038] Batch [3099]/[3760]	Speed: 68.772455 samples/sec	 accuracy=54.394153	 loss=1.921047	 lr=0.010000
Epoch[038] Batch [3149]/[3760]	Speed: 70.133914 samples/sec	 accuracy=54.370040	 loss=1.921643	 lr=0.010000
Epoch[038] Batch [3199]/[3760]	Speed: 69.269404 samples/sec	 accuracy=54.372070	 loss=1.922175	 lr=0.010000
Epoch[038] Batch [3249]/[3760]	Speed: 70.273078 samples/sec	 accuracy=54.338942	 loss=1.923314	 lr=0.010000
Epoch[038] Batch [3299]/[3760]	Speed: 69.359102 samples/sec	 accuracy=54.357008	 loss=1.922810	 lr=0.010000
Epoch[038] Batch [3349]/[3760]	Speed: 69.321935 samples/sec	 accuracy=54.364739	 loss=1.922700	 lr=0.010000
Epoch[038] Batch [3399]/[3760]	Speed: 71.037954 samples/sec	 accuracy=54.370404	 loss=1.922338	 lr=0.010000
Epoch[038] Batch [3449]/[3760]	Speed: 69.560562 samples/sec	 accuracy=54.348732	 loss=1.923726	 lr=0.010000
Epoch[038] Batch [3499]/[3760]	Speed: 69.434039 samples/sec	 accuracy=54.342411	 loss=1.924430	 lr=0.010000
Epoch[038] Batch [3549]/[3760]	Speed: 70.319566 samples/sec	 accuracy=54.362676	 loss=1.923990	 lr=0.010000
Epoch[038] Batch [3599]/[3760]	Speed: 69.625213 samples/sec	 accuracy=54.379774	 loss=1.923978	 lr=0.010000
Epoch[038] Batch [3649]/[3760]	Speed: 69.416161 samples/sec	 accuracy=54.388271	 loss=1.923853	 lr=0.010000
Epoch[038] Batch [3699]/[3760]	Speed: 70.679313 samples/sec	 accuracy=54.394426	 loss=1.923933	 lr=0.010000
Epoch[038] Batch [3749]/[3760]	Speed: 77.119337 samples/sec	 accuracy=54.387917	 loss=1.923888	 lr=0.010000
Batch [0049]/[0303]: acc-top1=53.625000 acc-top5=78.812500
Batch [0099]/[0303]: acc-top1=54.140625 acc-top5=78.906250
Batch [0149]/[0303]: acc-top1=54.739583 acc-top5=79.260417
Batch [0199]/[0303]: acc-top1=55.140625 acc-top5=79.406250
Batch [0249]/[0303]: acc-top1=55.275000 acc-top5=79.525000
Batch [0299]/[0303]: acc-top1=55.192708 acc-top5=79.171875
[Epoch 038] training: accuracy=54.379156	 loss=1.924286
[Epoch 038] speed: 69 samples/sec	time cost: 3755.097992
[Epoch 038] validation: acc-top1=55.151609 acc-top5=79.151196 loss=1.987875
Epoch[039] Batch [0049]/[3760]	Speed: 48.076783 samples/sec	 accuracy=55.000000	 loss=1.858462	 lr=0.010000
Epoch[039] Batch [0099]/[3760]	Speed: 69.533620 samples/sec	 accuracy=54.343750	 loss=1.884016	 lr=0.010000
Epoch[039] Batch [0149]/[3760]	Speed: 68.061985 samples/sec	 accuracy=54.906250	 loss=1.890047	 lr=0.010000
Epoch[039] Batch [0199]/[3760]	Speed: 69.395431 samples/sec	 accuracy=54.835938	 loss=1.899067	 lr=0.010000
Epoch[039] Batch [0249]/[3760]	Speed: 69.810515 samples/sec	 accuracy=54.918750	 loss=1.899660	 lr=0.010000
Epoch[039] Batch [0299]/[3760]	Speed: 70.578440 samples/sec	 accuracy=55.067708	 loss=1.895446	 lr=0.010000
Epoch[039] Batch [0349]/[3760]	Speed: 70.235290 samples/sec	 accuracy=54.946429	 loss=1.904490	 lr=0.010000
Epoch[039] Batch [0399]/[3760]	Speed: 69.405359 samples/sec	 accuracy=54.929688	 loss=1.900644	 lr=0.010000
Epoch[039] Batch [0449]/[3760]	Speed: 70.494513 samples/sec	 accuracy=54.815972	 loss=1.908493	 lr=0.010000
Epoch[039] Batch [0499]/[3760]	Speed: 70.940606 samples/sec	 accuracy=54.868750	 loss=1.906166	 lr=0.010000
Epoch[039] Batch [0549]/[3760]	Speed: 68.987217 samples/sec	 accuracy=54.974432	 loss=1.904392	 lr=0.010000
Epoch[039] Batch [0599]/[3760]	Speed: 69.957351 samples/sec	 accuracy=54.937500	 loss=1.906872	 lr=0.010000
Epoch[039] Batch [0649]/[3760]	Speed: 68.200597 samples/sec	 accuracy=54.908654	 loss=1.907324	 lr=0.010000
Epoch[039] Batch [0699]/[3760]	Speed: 70.782926 samples/sec	 accuracy=54.819196	 loss=1.910285	 lr=0.010000
Epoch[039] Batch [0749]/[3760]	Speed: 69.283123 samples/sec	 accuracy=54.735417	 loss=1.911905	 lr=0.010000
Epoch[039] Batch [0799]/[3760]	Speed: 68.647493 samples/sec	 accuracy=54.750000	 loss=1.911126	 lr=0.010000
Epoch[039] Batch [0849]/[3760]	Speed: 68.463018 samples/sec	 accuracy=54.886029	 loss=1.905157	 lr=0.010000
Epoch[039] Batch [0899]/[3760]	Speed: 70.012095 samples/sec	 accuracy=54.899306	 loss=1.906563	 lr=0.010000
Epoch[039] Batch [0949]/[3760]	Speed: 69.665729 samples/sec	 accuracy=54.909539	 loss=1.906786	 lr=0.010000
Epoch[039] Batch [0999]/[3760]	Speed: 69.108436 samples/sec	 accuracy=54.895313	 loss=1.906097	 lr=0.010000
Epoch[039] Batch [1049]/[3760]	Speed: 71.072482 samples/sec	 accuracy=54.883929	 loss=1.905272	 lr=0.010000
Epoch[039] Batch [1099]/[3760]	Speed: 68.290303 samples/sec	 accuracy=54.855114	 loss=1.905560	 lr=0.010000
Epoch[039] Batch [1149]/[3760]	Speed: 66.912456 samples/sec	 accuracy=54.846467	 loss=1.907437	 lr=0.010000
Epoch[039] Batch [1199]/[3760]	Speed: 72.756338 samples/sec	 accuracy=54.876302	 loss=1.907192	 lr=0.010000
Epoch[039] Batch [1249]/[3760]	Speed: 69.557440 samples/sec	 accuracy=54.910000	 loss=1.906148	 lr=0.010000
Epoch[039] Batch [1299]/[3760]	Speed: 70.293027 samples/sec	 accuracy=54.965144	 loss=1.903806	 lr=0.010000
Epoch[039] Batch [1349]/[3760]	Speed: 69.962458 samples/sec	 accuracy=54.978009	 loss=1.901596	 lr=0.010000
Epoch[039] Batch [1399]/[3760]	Speed: 67.503523 samples/sec	 accuracy=55.022321	 loss=1.900550	 lr=0.010000
Epoch[039] Batch [1449]/[3760]	Speed: 69.988585 samples/sec	 accuracy=54.971983	 loss=1.901894	 lr=0.010000
Epoch[039] Batch [1499]/[3760]	Speed: 70.122568 samples/sec	 accuracy=54.981250	 loss=1.902159	 lr=0.010000
Epoch[039] Batch [1549]/[3760]	Speed: 69.253251 samples/sec	 accuracy=54.941532	 loss=1.904174	 lr=0.010000
Epoch[039] Batch [1599]/[3760]	Speed: 70.227732 samples/sec	 accuracy=54.926758	 loss=1.903777	 lr=0.010000
Epoch[039] Batch [1649]/[3760]	Speed: 67.972095 samples/sec	 accuracy=54.914773	 loss=1.904956	 lr=0.010000
Epoch[039] Batch [1699]/[3760]	Speed: 70.815066 samples/sec	 accuracy=54.886029	 loss=1.905236	 lr=0.010000
Epoch[039] Batch [1749]/[3760]	Speed: 68.652008 samples/sec	 accuracy=54.858929	 loss=1.907751	 lr=0.010000
Epoch[039] Batch [1799]/[3760]	Speed: 69.895479 samples/sec	 accuracy=54.829861	 loss=1.907242	 lr=0.010000
Epoch[039] Batch [1849]/[3760]	Speed: 71.211686 samples/sec	 accuracy=54.871622	 loss=1.906799	 lr=0.010000
Epoch[039] Batch [1899]/[3760]	Speed: 69.024605 samples/sec	 accuracy=54.872533	 loss=1.907430	 lr=0.010000
Epoch[039] Batch [1949]/[3760]	Speed: 68.597456 samples/sec	 accuracy=54.868590	 loss=1.907471	 lr=0.010000
Epoch[039] Batch [1999]/[3760]	Speed: 70.061712 samples/sec	 accuracy=54.856250	 loss=1.908489	 lr=0.010000
Epoch[039] Batch [2049]/[3760]	Speed: 69.662846 samples/sec	 accuracy=54.838415	 loss=1.908430	 lr=0.010000
Epoch[039] Batch [2099]/[3760]	Speed: 68.607883 samples/sec	 accuracy=54.845982	 loss=1.908200	 lr=0.010000
Epoch[039] Batch [2149]/[3760]	Speed: 70.816197 samples/sec	 accuracy=54.806686	 loss=1.909492	 lr=0.010000
Epoch[039] Batch [2199]/[3760]	Speed: 68.318207 samples/sec	 accuracy=54.795455	 loss=1.909901	 lr=0.010000
Epoch[039] Batch [2249]/[3760]	Speed: 70.508615 samples/sec	 accuracy=54.809028	 loss=1.908868	 lr=0.010000
Epoch[039] Batch [2299]/[3760]	Speed: 69.367001 samples/sec	 accuracy=54.773098	 loss=1.910537	 lr=0.010000
Epoch[039] Batch [2349]/[3760]	Speed: 70.285280 samples/sec	 accuracy=54.774601	 loss=1.909853	 lr=0.010000
Epoch[039] Batch [2399]/[3760]	Speed: 68.584709 samples/sec	 accuracy=54.769531	 loss=1.910219	 lr=0.010000
Epoch[039] Batch [2449]/[3760]	Speed: 70.461894 samples/sec	 accuracy=54.759566	 loss=1.910443	 lr=0.010000
Epoch[039] Batch [2499]/[3760]	Speed: 68.949883 samples/sec	 accuracy=54.730000	 loss=1.911648	 lr=0.010000
Epoch[039] Batch [2549]/[3760]	Speed: 68.467906 samples/sec	 accuracy=54.733456	 loss=1.910163	 lr=0.010000
Epoch[039] Batch [2599]/[3760]	Speed: 70.625144 samples/sec	 accuracy=54.712139	 loss=1.911374	 lr=0.010000
Epoch[039] Batch [2649]/[3760]	Speed: 70.194267 samples/sec	 accuracy=54.738797	 loss=1.909998	 lr=0.010000
Epoch[039] Batch [2699]/[3760]	Speed: 69.224384 samples/sec	 accuracy=54.736690	 loss=1.909733	 lr=0.010000
Epoch[039] Batch [2749]/[3760]	Speed: 69.609925 samples/sec	 accuracy=54.720455	 loss=1.911208	 lr=0.010000
Epoch[039] Batch [2799]/[3760]	Speed: 70.343145 samples/sec	 accuracy=54.704799	 loss=1.912278	 lr=0.010000
Epoch[039] Batch [2849]/[3760]	Speed: 69.769416 samples/sec	 accuracy=54.733553	 loss=1.911969	 lr=0.010000
Epoch[039] Batch [2899]/[3760]	Speed: 69.607341 samples/sec	 accuracy=54.724677	 loss=1.912668	 lr=0.010000
Epoch[039] Batch [2949]/[3760]	Speed: 69.257567 samples/sec	 accuracy=54.730932	 loss=1.912380	 lr=0.010000
Epoch[039] Batch [2999]/[3760]	Speed: 68.612079 samples/sec	 accuracy=54.716667	 loss=1.912505	 lr=0.010000
Epoch[039] Batch [3049]/[3760]	Speed: 70.242961 samples/sec	 accuracy=54.736680	 loss=1.911849	 lr=0.010000
Epoch[039] Batch [3099]/[3760]	Speed: 69.933913 samples/sec	 accuracy=54.718246	 loss=1.912683	 lr=0.010000
Epoch[039] Batch [3149]/[3760]	Speed: 67.838289 samples/sec	 accuracy=54.689484	 loss=1.914422	 lr=0.010000
Epoch[039] Batch [3199]/[3760]	Speed: 71.398964 samples/sec	 accuracy=54.693359	 loss=1.914389	 lr=0.010000
Epoch[039] Batch [3249]/[3760]	Speed: 70.063379 samples/sec	 accuracy=54.680769	 loss=1.914685	 lr=0.010000
Epoch[039] Batch [3299]/[3760]	Speed: 68.228140 samples/sec	 accuracy=54.681345	 loss=1.915010	 lr=0.010000
Epoch[039] Batch [3349]/[3760]	Speed: 70.613322 samples/sec	 accuracy=54.670709	 loss=1.915837	 lr=0.010000
Epoch[039] Batch [3399]/[3760]	Speed: 69.121927 samples/sec	 accuracy=54.661305	 loss=1.916182	 lr=0.010000
Epoch[039] Batch [3449]/[3760]	Speed: 70.219974 samples/sec	 accuracy=54.663496	 loss=1.915925	 lr=0.010000
Epoch[039] Batch [3499]/[3760]	Speed: 70.984341 samples/sec	 accuracy=54.672768	 loss=1.915196	 lr=0.010000
Epoch[039] Batch [3549]/[3760]	Speed: 69.930705 samples/sec	 accuracy=54.649648	 loss=1.915984	 lr=0.010000
Epoch[039] Batch [3599]/[3760]	Speed: 68.643672 samples/sec	 accuracy=54.640625	 loss=1.916759	 lr=0.010000
Epoch[039] Batch [3649]/[3760]	Speed: 69.836770 samples/sec	 accuracy=54.646404	 loss=1.916279	 lr=0.010000
Epoch[039] Batch [3699]/[3760]	Speed: 68.308570 samples/sec	 accuracy=54.649916	 loss=1.916139	 lr=0.010000
Epoch[039] Batch [3749]/[3760]	Speed: 78.975231 samples/sec	 accuracy=54.652917	 loss=1.916445	 lr=0.001000
Batch [0049]/[0303]: acc-top1=53.343750 acc-top5=78.562500
Batch [0099]/[0303]: acc-top1=53.843750 acc-top5=78.578125
Batch [0149]/[0303]: acc-top1=54.812500 acc-top5=78.958333
Batch [0199]/[0303]: acc-top1=55.289062 acc-top5=79.148438
Batch [0249]/[0303]: acc-top1=55.606250 acc-top5=79.350000
Batch [0299]/[0303]: acc-top1=55.317708 acc-top5=79.151042
[Epoch 039] training: accuracy=54.643866	 loss=1.916899
[Epoch 039] speed: 69 samples/sec	time cost: 3759.962952
[Epoch 039] validation: acc-top1=55.301155 acc-top5=79.140883 loss=1.998273
Epoch[040] Batch [0049]/[3759]	Speed: 48.166652 samples/sec	 accuracy=54.625000	 loss=1.946110	 lr=0.001000
Epoch[040] Batch [0099]/[3759]	Speed: 68.838870 samples/sec	 accuracy=55.859375	 loss=1.869891	 lr=0.001000
Epoch[040] Batch [0149]/[3759]	Speed: 69.523954 samples/sec	 accuracy=56.718750	 loss=1.832492	 lr=0.001000
Epoch[040] Batch [0199]/[3759]	Speed: 70.077706 samples/sec	 accuracy=57.164062	 loss=1.808540	 lr=0.001000
Epoch[040] Batch [0249]/[3759]	Speed: 68.843700 samples/sec	 accuracy=57.393750	 loss=1.793640	 lr=0.001000
Epoch[040] Batch [0299]/[3759]	Speed: 68.079005 samples/sec	 accuracy=57.536458	 loss=1.783177	 lr=0.001000
Epoch[040] Batch [0349]/[3759]	Speed: 70.931840 samples/sec	 accuracy=57.651786	 loss=1.773427	 lr=0.001000
Epoch[040] Batch [0399]/[3759]	Speed: 70.433912 samples/sec	 accuracy=58.042969	 loss=1.757703	 lr=0.001000
Epoch[040] Batch [0449]/[3759]	Speed: 69.446575 samples/sec	 accuracy=58.211806	 loss=1.750251	 lr=0.001000
Epoch[040] Batch [0499]/[3759]	Speed: 70.397649 samples/sec	 accuracy=58.321875	 loss=1.741635	 lr=0.001000
Epoch[040] Batch [0549]/[3759]	Speed: 69.379666 samples/sec	 accuracy=58.491477	 loss=1.735225	 lr=0.001000
Epoch[040] Batch [0599]/[3759]	Speed: 68.850307 samples/sec	 accuracy=58.638021	 loss=1.731582	 lr=0.001000
Epoch[040] Batch [0649]/[3759]	Speed: 69.882156 samples/sec	 accuracy=58.723558	 loss=1.724201	 lr=0.001000
Epoch[040] Batch [0699]/[3759]	Speed: 69.288828 samples/sec	 accuracy=58.794643	 loss=1.720775	 lr=0.001000
Epoch[040] Batch [0749]/[3759]	Speed: 70.522169 samples/sec	 accuracy=58.860417	 loss=1.718226	 lr=0.001000
Epoch[040] Batch [0799]/[3759]	Speed: 69.613044 samples/sec	 accuracy=58.951172	 loss=1.715179	 lr=0.001000
Epoch[040] Batch [0849]/[3759]	Speed: 68.430425 samples/sec	 accuracy=59.077206	 loss=1.712080	 lr=0.001000
Epoch[040] Batch [0899]/[3759]	Speed: 69.443143 samples/sec	 accuracy=59.161458	 loss=1.707436	 lr=0.001000
Epoch[040] Batch [0949]/[3759]	Speed: 70.242954 samples/sec	 accuracy=59.268092	 loss=1.703847	 lr=0.001000
Epoch[040] Batch [0999]/[3759]	Speed: 70.428855 samples/sec	 accuracy=59.271875	 loss=1.700437	 lr=0.001000
Epoch[040] Batch [1049]/[3759]	Speed: 69.451622 samples/sec	 accuracy=59.312500	 loss=1.697996	 lr=0.001000
Epoch[040] Batch [1099]/[3759]	Speed: 67.771903 samples/sec	 accuracy=59.345170	 loss=1.694919	 lr=0.001000
Epoch[040] Batch [1149]/[3759]	Speed: 70.674334 samples/sec	 accuracy=59.384511	 loss=1.693553	 lr=0.001000
Epoch[040] Batch [1199]/[3759]	Speed: 70.726106 samples/sec	 accuracy=59.453125	 loss=1.690294	 lr=0.001000
Epoch[040] Batch [1249]/[3759]	Speed: 68.641409 samples/sec	 accuracy=59.520000	 loss=1.687698	 lr=0.001000
Epoch[040] Batch [1299]/[3759]	Speed: 69.457880 samples/sec	 accuracy=59.592548	 loss=1.687110	 lr=0.001000
Epoch[040] Batch [1349]/[3759]	Speed: 70.752808 samples/sec	 accuracy=59.630787	 loss=1.684434	 lr=0.001000
Epoch[040] Batch [1399]/[3759]	Speed: 69.734168 samples/sec	 accuracy=59.678571	 loss=1.680936	 lr=0.001000
Epoch[040] Batch [1449]/[3759]	Speed: 70.941416 samples/sec	 accuracy=59.698276	 loss=1.680433	 lr=0.001000
Epoch[040] Batch [1499]/[3759]	Speed: 70.082745 samples/sec	 accuracy=59.746875	 loss=1.678999	 lr=0.001000
Epoch[040] Batch [1549]/[3759]	Speed: 69.317644 samples/sec	 accuracy=59.791331	 loss=1.677755	 lr=0.001000
Epoch[040] Batch [1599]/[3759]	Speed: 67.751588 samples/sec	 accuracy=59.811523	 loss=1.676774	 lr=0.001000
Epoch[040] Batch [1649]/[3759]	Speed: 71.451519 samples/sec	 accuracy=59.877841	 loss=1.675133	 lr=0.001000
Epoch[040] Batch [1699]/[3759]	Speed: 70.123733 samples/sec	 accuracy=59.922794	 loss=1.673321	 lr=0.001000
Epoch[040] Batch [1749]/[3759]	Speed: 68.350290 samples/sec	 accuracy=59.950000	 loss=1.671298	 lr=0.001000
Epoch[040] Batch [1799]/[3759]	Speed: 70.194099 samples/sec	 accuracy=60.001736	 loss=1.668792	 lr=0.001000
Epoch[040] Batch [1849]/[3759]	Speed: 69.809604 samples/sec	 accuracy=60.020270	 loss=1.667357	 lr=0.001000
Epoch[040] Batch [1899]/[3759]	Speed: 69.195448 samples/sec	 accuracy=60.046875	 loss=1.666059	 lr=0.001000
Epoch[040] Batch [1949]/[3759]	Speed: 69.168836 samples/sec	 accuracy=60.044071	 loss=1.666058	 lr=0.001000
Epoch[040] Batch [1999]/[3759]	Speed: 70.267612 samples/sec	 accuracy=60.051563	 loss=1.665014	 lr=0.001000
Epoch[040] Batch [2049]/[3759]	Speed: 70.139950 samples/sec	 accuracy=60.099085	 loss=1.662841	 lr=0.001000
Epoch[040] Batch [2099]/[3759]	Speed: 70.022670 samples/sec	 accuracy=60.125744	 loss=1.662572	 lr=0.001000
Epoch[040] Batch [2149]/[3759]	Speed: 69.505347 samples/sec	 accuracy=60.154797	 loss=1.661083	 lr=0.001000
Epoch[040] Batch [2199]/[3759]	Speed: 69.660725 samples/sec	 accuracy=60.191761	 loss=1.659318	 lr=0.001000
Epoch[040] Batch [2249]/[3759]	Speed: 70.544664 samples/sec	 accuracy=60.222917	 loss=1.658830	 lr=0.001000
Epoch[040] Batch [2299]/[3759]	Speed: 68.588715 samples/sec	 accuracy=60.260190	 loss=1.657394	 lr=0.001000
Epoch[040] Batch [2349]/[3759]	Speed: 69.502735 samples/sec	 accuracy=60.299202	 loss=1.656398	 lr=0.001000
Epoch[040] Batch [2399]/[3759]	Speed: 70.975580 samples/sec	 accuracy=60.327474	 loss=1.654690	 lr=0.001000
Epoch[040] Batch [2449]/[3759]	Speed: 70.890722 samples/sec	 accuracy=60.313776	 loss=1.655036	 lr=0.001000
Epoch[040] Batch [2499]/[3759]	Speed: 68.428503 samples/sec	 accuracy=60.340000	 loss=1.653727	 lr=0.001000
Epoch[040] Batch [2549]/[3759]	Speed: 70.791012 samples/sec	 accuracy=60.362132	 loss=1.652336	 lr=0.001000
Epoch[040] Batch [2599]/[3759]	Speed: 70.838665 samples/sec	 accuracy=60.387019	 loss=1.650913	 lr=0.001000
Epoch[040] Batch [2649]/[3759]	Speed: 69.196989 samples/sec	 accuracy=60.392099	 loss=1.650430	 lr=0.001000
Epoch[040] Batch [2699]/[3759]	Speed: 68.555191 samples/sec	 accuracy=60.377894	 loss=1.650675	 lr=0.001000
Epoch[040] Batch [2749]/[3759]	Speed: 69.708302 samples/sec	 accuracy=60.405114	 loss=1.649135	 lr=0.001000
Epoch[040] Batch [2799]/[3759]	Speed: 69.906024 samples/sec	 accuracy=60.449219	 loss=1.646939	 lr=0.001000
Epoch[040] Batch [2849]/[3759]	Speed: 70.185052 samples/sec	 accuracy=60.476974	 loss=1.645644	 lr=0.001000
Epoch[040] Batch [2899]/[3759]	Speed: 70.103445 samples/sec	 accuracy=60.488685	 loss=1.645272	 lr=0.001000
Epoch[040] Batch [2949]/[3759]	Speed: 68.868958 samples/sec	 accuracy=60.508475	 loss=1.645160	 lr=0.001000
Epoch[040] Batch [2999]/[3759]	Speed: 69.637975 samples/sec	 accuracy=60.510937	 loss=1.645150	 lr=0.001000
Epoch[040] Batch [3049]/[3759]	Speed: 70.414844 samples/sec	 accuracy=60.510758	 loss=1.644708	 lr=0.001000
Epoch[040] Batch [3099]/[3759]	Speed: 70.308213 samples/sec	 accuracy=60.517137	 loss=1.643819	 lr=0.001000
Epoch[040] Batch [3149]/[3759]	Speed: 70.599842 samples/sec	 accuracy=60.542163	 loss=1.642721	 lr=0.001000
Epoch[040] Batch [3199]/[3759]	Speed: 70.081762 samples/sec	 accuracy=60.562500	 loss=1.641570	 lr=0.001000
Epoch[040] Batch [3249]/[3759]	Speed: 68.121276 samples/sec	 accuracy=60.575962	 loss=1.641103	 lr=0.001000
Epoch[040] Batch [3299]/[3759]	Speed: 69.655907 samples/sec	 accuracy=60.602273	 loss=1.639644	 lr=0.001000
Epoch[040] Batch [3349]/[3759]	Speed: 69.740598 samples/sec	 accuracy=60.601679	 loss=1.639419	 lr=0.001000
Epoch[040] Batch [3399]/[3759]	Speed: 70.021356 samples/sec	 accuracy=60.619945	 loss=1.638587	 lr=0.001000
Epoch[040] Batch [3449]/[3759]	Speed: 70.341798 samples/sec	 accuracy=60.641304	 loss=1.637649	 lr=0.001000
Epoch[040] Batch [3499]/[3759]	Speed: 68.995621 samples/sec	 accuracy=60.682143	 loss=1.636221	 lr=0.001000
Epoch[040] Batch [3549]/[3759]	Speed: 69.537767 samples/sec	 accuracy=60.672975	 loss=1.636126	 lr=0.001000
Epoch[040] Batch [3599]/[3759]	Speed: 69.160736 samples/sec	 accuracy=60.685764	 loss=1.635553	 lr=0.001000
Epoch[040] Batch [3649]/[3759]	Speed: 70.054738 samples/sec	 accuracy=60.680223	 loss=1.635675	 lr=0.001000
Epoch[040] Batch [3699]/[3759]	Speed: 70.532035 samples/sec	 accuracy=60.693412	 loss=1.635661	 lr=0.001000
Epoch[040] Batch [3749]/[3759]	Speed: 78.247883 samples/sec	 accuracy=60.711250	 loss=1.635068	 lr=0.001000
Batch [0049]/[0303]: acc-top1=59.625000 acc-top5=82.062500
Batch [0099]/[0303]: acc-top1=59.765625 acc-top5=82.109375
Batch [0149]/[0303]: acc-top1=60.656250 acc-top5=82.729167
Batch [0199]/[0303]: acc-top1=60.968750 acc-top5=83.117188
Batch [0249]/[0303]: acc-top1=61.000000 acc-top5=83.237500
Batch [0299]/[0303]: acc-top1=60.744792 acc-top5=83.083333
[Epoch 040] training: accuracy=60.713454	 loss=1.635091
[Epoch 040] speed: 69 samples/sec	time cost: 3752.395788
[Epoch 040] validation: acc-top1=60.746700 acc-top5=83.080652 loss=1.703602
Epoch[041] Batch [0049]/[3760]	Speed: 47.982814 samples/sec	 accuracy=62.062500	 loss=1.616788	 lr=0.001000
Epoch[041] Batch [0099]/[3760]	Speed: 69.946791 samples/sec	 accuracy=62.437500	 loss=1.567000	 lr=0.001000
Epoch[041] Batch [0149]/[3760]	Speed: 68.835957 samples/sec	 accuracy=61.781250	 loss=1.576514	 lr=0.001000
Epoch[041] Batch [0199]/[3760]	Speed: 70.397876 samples/sec	 accuracy=61.812500	 loss=1.577453	 lr=0.001000
Epoch[041] Batch [0249]/[3760]	Speed: 70.651914 samples/sec	 accuracy=62.012500	 loss=1.576993	 lr=0.001000
Epoch[041] Batch [0299]/[3760]	Speed: 70.231577 samples/sec	 accuracy=62.015625	 loss=1.571509	 lr=0.001000
Epoch[041] Batch [0349]/[3760]	Speed: 67.937697 samples/sec	 accuracy=62.000000	 loss=1.571061	 lr=0.001000
Epoch[041] Batch [0399]/[3760]	Speed: 70.723858 samples/sec	 accuracy=61.914062	 loss=1.569577	 lr=0.001000
Epoch[041] Batch [0449]/[3760]	Speed: 70.596886 samples/sec	 accuracy=61.923611	 loss=1.569243	 lr=0.001000
Epoch[041] Batch [0499]/[3760]	Speed: 68.447859 samples/sec	 accuracy=61.946875	 loss=1.572950	 lr=0.001000
Epoch[041] Batch [0549]/[3760]	Speed: 69.690770 samples/sec	 accuracy=61.943182	 loss=1.575255	 lr=0.001000
Epoch[041] Batch [0599]/[3760]	Speed: 68.289235 samples/sec	 accuracy=62.101562	 loss=1.568566	 lr=0.001000
Epoch[041] Batch [0649]/[3760]	Speed: 69.493802 samples/sec	 accuracy=62.108173	 loss=1.567924	 lr=0.001000
Epoch[041] Batch [0699]/[3760]	Speed: 69.964534 samples/sec	 accuracy=62.102679	 loss=1.567953	 lr=0.001000
Epoch[041] Batch [0749]/[3760]	Speed: 70.967387 samples/sec	 accuracy=62.027083	 loss=1.574353	 lr=0.001000
Epoch[041] Batch [0799]/[3760]	Speed: 66.611692 samples/sec	 accuracy=61.992187	 loss=1.576480	 lr=0.001000
Epoch[041] Batch [0849]/[3760]	Speed: 71.348545 samples/sec	 accuracy=61.920956	 loss=1.576287	 lr=0.001000
Epoch[041] Batch [0899]/[3760]	Speed: 69.587686 samples/sec	 accuracy=61.859375	 loss=1.578002	 lr=0.001000
Epoch[041] Batch [0949]/[3760]	Speed: 70.186095 samples/sec	 accuracy=61.807566	 loss=1.578123	 lr=0.001000
Epoch[041] Batch [0999]/[3760]	Speed: 69.312850 samples/sec	 accuracy=61.859375	 loss=1.577407	 lr=0.001000
Epoch[041] Batch [1049]/[3760]	Speed: 70.588000 samples/sec	 accuracy=61.906250	 loss=1.577258	 lr=0.001000
Epoch[041] Batch [1099]/[3760]	Speed: 70.179776 samples/sec	 accuracy=61.886364	 loss=1.577423	 lr=0.001000
Epoch[041] Batch [1149]/[3760]	Speed: 69.776344 samples/sec	 accuracy=61.955163	 loss=1.574685	 lr=0.001000
Epoch[041] Batch [1199]/[3760]	Speed: 69.464615 samples/sec	 accuracy=61.953125	 loss=1.574626	 lr=0.001000
Epoch[041] Batch [1249]/[3760]	Speed: 70.202761 samples/sec	 accuracy=61.956250	 loss=1.575004	 lr=0.001000
Epoch[041] Batch [1299]/[3760]	Speed: 68.110541 samples/sec	 accuracy=61.926683	 loss=1.576285	 lr=0.001000
Epoch[041] Batch [1349]/[3760]	Speed: 71.899077 samples/sec	 accuracy=61.942130	 loss=1.576253	 lr=0.001000
Epoch[041] Batch [1399]/[3760]	Speed: 69.081367 samples/sec	 accuracy=61.938616	 loss=1.576664	 lr=0.001000
Epoch[041] Batch [1449]/[3760]	Speed: 69.032439 samples/sec	 accuracy=61.922414	 loss=1.576728	 lr=0.001000
Epoch[041] Batch [1499]/[3760]	Speed: 69.520771 samples/sec	 accuracy=61.903125	 loss=1.577399	 lr=0.001000
Epoch[041] Batch [1549]/[3760]	Speed: 70.698106 samples/sec	 accuracy=61.929435	 loss=1.576450	 lr=0.001000
Epoch[041] Batch [1599]/[3760]	Speed: 69.672603 samples/sec	 accuracy=61.921875	 loss=1.576795	 lr=0.001000
Epoch[041] Batch [1649]/[3760]	Speed: 69.821481 samples/sec	 accuracy=61.905303	 loss=1.575889	 lr=0.001000
Epoch[041] Batch [1699]/[3760]	Speed: 69.635953 samples/sec	 accuracy=61.872243	 loss=1.577074	 lr=0.001000
Epoch[041] Batch [1749]/[3760]	Speed: 69.223233 samples/sec	 accuracy=61.844643	 loss=1.577768	 lr=0.001000
Epoch[041] Batch [1799]/[3760]	Speed: 70.408263 samples/sec	 accuracy=61.835069	 loss=1.578083	 lr=0.001000
Epoch[041] Batch [1849]/[3760]	Speed: 68.680858 samples/sec	 accuracy=61.828547	 loss=1.578628	 lr=0.001000
Epoch[041] Batch [1899]/[3760]	Speed: 68.467976 samples/sec	 accuracy=61.875000	 loss=1.576913	 lr=0.001000
Epoch[041] Batch [1949]/[3760]	Speed: 70.432452 samples/sec	 accuracy=61.883814	 loss=1.577152	 lr=0.001000
Epoch[041] Batch [1999]/[3760]	Speed: 69.376536 samples/sec	 accuracy=61.871094	 loss=1.577089	 lr=0.001000
Epoch[041] Batch [2049]/[3760]	Speed: 70.971903 samples/sec	 accuracy=61.868902	 loss=1.576784	 lr=0.001000
Epoch[041] Batch [2099]/[3760]	Speed: 69.051789 samples/sec	 accuracy=61.849702	 loss=1.576712	 lr=0.001000
Epoch[041] Batch [2149]/[3760]	Speed: 69.646925 samples/sec	 accuracy=61.836483	 loss=1.576585	 lr=0.001000
Epoch[041] Batch [2199]/[3760]	Speed: 69.862119 samples/sec	 accuracy=61.855824	 loss=1.575845	 lr=0.001000
Epoch[041] Batch [2249]/[3760]	Speed: 69.847016 samples/sec	 accuracy=61.874306	 loss=1.575756	 lr=0.001000
Epoch[041] Batch [2299]/[3760]	Speed: 69.677561 samples/sec	 accuracy=61.908967	 loss=1.575086	 lr=0.001000
Epoch[041] Batch [2349]/[3760]	Speed: 69.417813 samples/sec	 accuracy=61.901596	 loss=1.575027	 lr=0.001000
Epoch[041] Batch [2399]/[3760]	Speed: 67.770440 samples/sec	 accuracy=61.915365	 loss=1.575219	 lr=0.001000
Epoch[041] Batch [2449]/[3760]	Speed: 71.434048 samples/sec	 accuracy=61.904337	 loss=1.575385	 lr=0.001000
Epoch[041] Batch [2499]/[3760]	Speed: 69.408366 samples/sec	 accuracy=61.884375	 loss=1.575537	 lr=0.001000
Epoch[041] Batch [2549]/[3760]	Speed: 69.846969 samples/sec	 accuracy=61.881740	 loss=1.575774	 lr=0.001000
Epoch[041] Batch [2599]/[3760]	Speed: 69.486795 samples/sec	 accuracy=61.870192	 loss=1.576240	 lr=0.001000
Epoch[041] Batch [2649]/[3760]	Speed: 68.297742 samples/sec	 accuracy=61.886792	 loss=1.575614	 lr=0.001000
Epoch[041] Batch [2699]/[3760]	Speed: 69.587593 samples/sec	 accuracy=61.894676	 loss=1.574857	 lr=0.001000
Epoch[041] Batch [2749]/[3760]	Speed: 71.598030 samples/sec	 accuracy=61.913636	 loss=1.573963	 lr=0.001000
Epoch[041] Batch [2799]/[3760]	Speed: 69.725023 samples/sec	 accuracy=61.916295	 loss=1.573741	 lr=0.001000
Epoch[041] Batch [2849]/[3760]	Speed: 64.435254 samples/sec	 accuracy=61.930373	 loss=1.573818	 lr=0.001000
Epoch[041] Batch [2899]/[3760]	Speed: 72.629167 samples/sec	 accuracy=61.941272	 loss=1.573838	 lr=0.001000
Epoch[041] Batch [2949]/[3760]	Speed: 67.715341 samples/sec	 accuracy=61.929555	 loss=1.574014	 lr=0.001000
Epoch[041] Batch [2999]/[3760]	Speed: 72.235833 samples/sec	 accuracy=61.941667	 loss=1.573756	 lr=0.001000
Epoch[041] Batch [3049]/[3760]	Speed: 68.460431 samples/sec	 accuracy=61.973361	 loss=1.571957	 lr=0.001000
Epoch[041] Batch [3099]/[3760]	Speed: 69.338946 samples/sec	 accuracy=61.953125	 loss=1.572258	 lr=0.001000
Epoch[041] Batch [3149]/[3760]	Speed: 71.395372 samples/sec	 accuracy=61.977679	 loss=1.571830	 lr=0.001000
Epoch[041] Batch [3199]/[3760]	Speed: 66.543167 samples/sec	 accuracy=61.997559	 loss=1.570882	 lr=0.001000
Epoch[041] Batch [3249]/[3760]	Speed: 71.156306 samples/sec	 accuracy=61.994231	 loss=1.570981	 lr=0.001000
Epoch[041] Batch [3299]/[3760]	Speed: 69.484749 samples/sec	 accuracy=62.022727	 loss=1.569621	 lr=0.001000
Epoch[041] Batch [3349]/[3760]	Speed: 68.964192 samples/sec	 accuracy=62.020989	 loss=1.569568	 lr=0.001000
Epoch[041] Batch [3399]/[3760]	Speed: 70.887852 samples/sec	 accuracy=62.027574	 loss=1.569425	 lr=0.001000
Epoch[041] Batch [3449]/[3760]	Speed: 69.336944 samples/sec	 accuracy=62.027627	 loss=1.569613	 lr=0.001000
Epoch[041] Batch [3499]/[3760]	Speed: 71.367698 samples/sec	 accuracy=62.033482	 loss=1.569171	 lr=0.001000
Epoch[041] Batch [3549]/[3760]	Speed: 68.882348 samples/sec	 accuracy=62.036972	 loss=1.569102	 lr=0.001000
Epoch[041] Batch [3599]/[3760]	Speed: 69.470841 samples/sec	 accuracy=62.050347	 loss=1.568199	 lr=0.001000
Epoch[041] Batch [3649]/[3760]	Speed: 67.161368 samples/sec	 accuracy=62.036387	 loss=1.567933	 lr=0.001000
Epoch[041] Batch [3699]/[3760]	Speed: 70.779183 samples/sec	 accuracy=62.055743	 loss=1.567056	 lr=0.001000
Epoch[041] Batch [3749]/[3760]	Speed: 76.965538 samples/sec	 accuracy=62.035833	 loss=1.567663	 lr=0.001000
Batch [0049]/[0303]: acc-top1=59.375000 acc-top5=82.593750
Batch [0099]/[0303]: acc-top1=59.984375 acc-top5=82.531250
Batch [0149]/[0303]: acc-top1=60.791667 acc-top5=83.041667
Batch [0199]/[0303]: acc-top1=61.101562 acc-top5=83.296875
Batch [0249]/[0303]: acc-top1=61.343750 acc-top5=83.487500
Batch [0299]/[0303]: acc-top1=61.255208 acc-top5=83.333333
[Epoch 041] training: accuracy=62.036652	 loss=1.567673
[Epoch 041] speed: 69 samples/sec	time cost: 3760.317774
[Epoch 041] validation: acc-top1=61.252063 acc-top5=83.338490 loss=1.685874
Epoch[042] Batch [0049]/[3760]	Speed: 48.618353 samples/sec	 accuracy=61.250000	 loss=1.578729	 lr=0.001000
Epoch[042] Batch [0099]/[3760]	Speed: 68.937291 samples/sec	 accuracy=61.796875	 loss=1.553123	 lr=0.001000
Epoch[042] Batch [0149]/[3760]	Speed: 68.448680 samples/sec	 accuracy=61.791667	 loss=1.556224	 lr=0.001000
Epoch[042] Batch [0199]/[3760]	Speed: 70.085066 samples/sec	 accuracy=62.273437	 loss=1.542781	 lr=0.001000
Epoch[042] Batch [0249]/[3760]	Speed: 69.683718 samples/sec	 accuracy=62.381250	 loss=1.541038	 lr=0.001000
Epoch[042] Batch [0299]/[3760]	Speed: 69.055825 samples/sec	 accuracy=62.640625	 loss=1.530655	 lr=0.001000
Epoch[042] Batch [0349]/[3760]	Speed: 69.416444 samples/sec	 accuracy=62.566964	 loss=1.532099	 lr=0.001000
Epoch[042] Batch [0399]/[3760]	Speed: 70.197637 samples/sec	 accuracy=62.488281	 loss=1.532159	 lr=0.001000
Epoch[042] Batch [0449]/[3760]	Speed: 66.696948 samples/sec	 accuracy=62.611111	 loss=1.532409	 lr=0.001000
Epoch[042] Batch [0499]/[3760]	Speed: 72.224244 samples/sec	 accuracy=62.706250	 loss=1.529212	 lr=0.001000
Epoch[042] Batch [0549]/[3760]	Speed: 68.892305 samples/sec	 accuracy=62.752841	 loss=1.525858	 lr=0.001000
Epoch[042] Batch [0599]/[3760]	Speed: 69.606375 samples/sec	 accuracy=62.729167	 loss=1.528172	 lr=0.001000
Epoch[042] Batch [0649]/[3760]	Speed: 70.046525 samples/sec	 accuracy=62.701923	 loss=1.528143	 lr=0.001000
Epoch[042] Batch [0699]/[3760]	Speed: 69.568217 samples/sec	 accuracy=62.720982	 loss=1.528172	 lr=0.001000
Epoch[042] Batch [0749]/[3760]	Speed: 69.500014 samples/sec	 accuracy=62.785417	 loss=1.529121	 lr=0.001000
Epoch[042] Batch [0799]/[3760]	Speed: 68.121558 samples/sec	 accuracy=62.830078	 loss=1.527016	 lr=0.001000
Epoch[042] Batch [0849]/[3760]	Speed: 69.800507 samples/sec	 accuracy=62.792279	 loss=1.528100	 lr=0.001000
Epoch[042] Batch [0899]/[3760]	Speed: 69.239915 samples/sec	 accuracy=62.776042	 loss=1.526360	 lr=0.001000
Epoch[042] Batch [0949]/[3760]	Speed: 68.257095 samples/sec	 accuracy=62.805921	 loss=1.526310	 lr=0.001000
Epoch[042] Batch [0999]/[3760]	Speed: 68.934080 samples/sec	 accuracy=62.835938	 loss=1.525468	 lr=0.001000
Epoch[042] Batch [1049]/[3760]	Speed: 70.784999 samples/sec	 accuracy=62.837798	 loss=1.526487	 lr=0.001000
Epoch[042] Batch [1099]/[3760]	Speed: 68.434547 samples/sec	 accuracy=62.879261	 loss=1.524185	 lr=0.001000
Epoch[042] Batch [1149]/[3760]	Speed: 68.179262 samples/sec	 accuracy=62.894022	 loss=1.524378	 lr=0.001000
Epoch[042] Batch [1199]/[3760]	Speed: 70.286030 samples/sec	 accuracy=62.880208	 loss=1.525190	 lr=0.001000
Epoch[042] Batch [1249]/[3760]	Speed: 69.239457 samples/sec	 accuracy=62.887500	 loss=1.525742	 lr=0.001000
Epoch[042] Batch [1299]/[3760]	Speed: 70.574053 samples/sec	 accuracy=62.867788	 loss=1.526529	 lr=0.001000
Epoch[042] Batch [1349]/[3760]	Speed: 69.682160 samples/sec	 accuracy=62.918981	 loss=1.525473	 lr=0.001000
Epoch[042] Batch [1399]/[3760]	Speed: 68.641288 samples/sec	 accuracy=62.908482	 loss=1.525525	 lr=0.001000
Epoch[042] Batch [1449]/[3760]	Speed: 68.797147 samples/sec	 accuracy=62.927802	 loss=1.525577	 lr=0.001000
Epoch[042] Batch [1499]/[3760]	Speed: 70.507274 samples/sec	 accuracy=62.918750	 loss=1.526196	 lr=0.001000
Epoch[042] Batch [1549]/[3760]	Speed: 70.768832 samples/sec	 accuracy=62.936492	 loss=1.525832	 lr=0.001000
Epoch[042] Batch [1599]/[3760]	Speed: 68.102756 samples/sec	 accuracy=62.910156	 loss=1.526091	 lr=0.001000
Epoch[042] Batch [1649]/[3760]	Speed: 71.866147 samples/sec	 accuracy=62.923295	 loss=1.525412	 lr=0.001000
Epoch[042] Batch [1699]/[3760]	Speed: 68.693249 samples/sec	 accuracy=62.909926	 loss=1.525461	 lr=0.001000
Epoch[042] Batch [1749]/[3760]	Speed: 69.325489 samples/sec	 accuracy=62.841964	 loss=1.527738	 lr=0.001000
Epoch[042] Batch [1799]/[3760]	Speed: 69.721550 samples/sec	 accuracy=62.866319	 loss=1.526962	 lr=0.001000
Epoch[042] Batch [1849]/[3760]	Speed: 69.562650 samples/sec	 accuracy=62.849662	 loss=1.526993	 lr=0.001000
Epoch[042] Batch [1899]/[3760]	Speed: 65.365790 samples/sec	 accuracy=62.823191	 loss=1.527091	 lr=0.001000
Epoch[042] Batch [1949]/[3760]	Speed: 73.893296 samples/sec	 accuracy=62.827724	 loss=1.527635	 lr=0.001000
Epoch[042] Batch [1999]/[3760]	Speed: 70.128833 samples/sec	 accuracy=62.845312	 loss=1.526975	 lr=0.001000
Epoch[042] Batch [2049]/[3760]	Speed: 69.136075 samples/sec	 accuracy=62.842226	 loss=1.527280	 lr=0.001000
Epoch[042] Batch [2099]/[3760]	Speed: 69.248055 samples/sec	 accuracy=62.841518	 loss=1.527590	 lr=0.001000
Epoch[042] Batch [2149]/[3760]	Speed: 69.743372 samples/sec	 accuracy=62.811773	 loss=1.529317	 lr=0.001000
Epoch[042] Batch [2199]/[3760]	Speed: 69.253360 samples/sec	 accuracy=62.830966	 loss=1.528585	 lr=0.001000
Epoch[042] Batch [2249]/[3760]	Speed: 70.138309 samples/sec	 accuracy=62.788889	 loss=1.530396	 lr=0.001000
Epoch[042] Batch [2299]/[3760]	Speed: 67.625616 samples/sec	 accuracy=62.811821	 loss=1.530266	 lr=0.001000
Epoch[042] Batch [2349]/[3760]	Speed: 69.836812 samples/sec	 accuracy=62.819814	 loss=1.529773	 lr=0.001000
Epoch[042] Batch [2399]/[3760]	Speed: 70.967642 samples/sec	 accuracy=62.822917	 loss=1.529679	 lr=0.001000
Epoch[042] Batch [2449]/[3760]	Speed: 69.728945 samples/sec	 accuracy=62.804847	 loss=1.530336	 lr=0.001000
Epoch[042] Batch [2499]/[3760]	Speed: 69.974414 samples/sec	 accuracy=62.791875	 loss=1.531159	 lr=0.001000
Epoch[042] Batch [2549]/[3760]	Speed: 65.645332 samples/sec	 accuracy=62.802083	 loss=1.530992	 lr=0.001000
Epoch[042] Batch [2599]/[3760]	Speed: 72.600699 samples/sec	 accuracy=62.806490	 loss=1.531735	 lr=0.001000
Epoch[042] Batch [2649]/[3760]	Speed: 71.112605 samples/sec	 accuracy=62.790684	 loss=1.532222	 lr=0.001000
Epoch[042] Batch [2699]/[3760]	Speed: 69.755363 samples/sec	 accuracy=62.771412	 loss=1.532816	 lr=0.001000
Epoch[042] Batch [2749]/[3760]	Speed: 67.344397 samples/sec	 accuracy=62.782386	 loss=1.533132	 lr=0.001000
Epoch[042] Batch [2799]/[3760]	Speed: 68.583301 samples/sec	 accuracy=62.770647	 loss=1.533522	 lr=0.001000
Epoch[042] Batch [2849]/[3760]	Speed: 72.053856 samples/sec	 accuracy=62.777412	 loss=1.534126	 lr=0.001000
Epoch[042] Batch [2899]/[3760]	Speed: 69.408274 samples/sec	 accuracy=62.779634	 loss=1.533949	 lr=0.001000
Epoch[042] Batch [2949]/[3760]	Speed: 70.305921 samples/sec	 accuracy=62.756356	 loss=1.534793	 lr=0.001000
Epoch[042] Batch [2999]/[3760]	Speed: 69.491230 samples/sec	 accuracy=62.746354	 loss=1.534644	 lr=0.001000
Epoch[042] Batch [3049]/[3760]	Speed: 69.339886 samples/sec	 accuracy=62.764344	 loss=1.534530	 lr=0.001000
Epoch[042] Batch [3099]/[3760]	Speed: 70.146419 samples/sec	 accuracy=62.775202	 loss=1.534453	 lr=0.001000
Epoch[042] Batch [3149]/[3760]	Speed: 70.339125 samples/sec	 accuracy=62.802579	 loss=1.534068	 lr=0.001000
Epoch[042] Batch [3199]/[3760]	Speed: 68.759004 samples/sec	 accuracy=62.797852	 loss=1.534732	 lr=0.001000
Epoch[042] Batch [3249]/[3760]	Speed: 71.189751 samples/sec	 accuracy=62.801442	 loss=1.534806	 lr=0.001000
Epoch[042] Batch [3299]/[3760]	Speed: 69.074900 samples/sec	 accuracy=62.800663	 loss=1.535416	 lr=0.001000
Epoch[042] Batch [3349]/[3760]	Speed: 68.935966 samples/sec	 accuracy=62.821828	 loss=1.534725	 lr=0.001000
Epoch[042] Batch [3399]/[3760]	Speed: 70.894798 samples/sec	 accuracy=62.809743	 loss=1.535589	 lr=0.001000
Epoch[042] Batch [3449]/[3760]	Speed: 68.394195 samples/sec	 accuracy=62.808877	 loss=1.535294	 lr=0.001000
Epoch[042] Batch [3499]/[3760]	Speed: 69.148218 samples/sec	 accuracy=62.806250	 loss=1.535297	 lr=0.001000
Epoch[042] Batch [3549]/[3760]	Speed: 70.824113 samples/sec	 accuracy=62.812940	 loss=1.534883	 lr=0.001000
Epoch[042] Batch [3599]/[3760]	Speed: 70.176438 samples/sec	 accuracy=62.815538	 loss=1.534541	 lr=0.001000
Epoch[042] Batch [3649]/[3760]	Speed: 68.568792 samples/sec	 accuracy=62.793664	 loss=1.534834	 lr=0.001000
Epoch[042] Batch [3699]/[3760]	Speed: 69.866022 samples/sec	 accuracy=62.797720	 loss=1.534517	 lr=0.001000
Epoch[042] Batch [3749]/[3760]	Speed: 77.545800 samples/sec	 accuracy=62.805833	 loss=1.534735	 lr=0.001000
Batch [0049]/[0303]: acc-top1=60.031250 acc-top5=82.718750
Batch [0099]/[0303]: acc-top1=59.984375 acc-top5=82.765625
Batch [0149]/[0303]: acc-top1=60.885417 acc-top5=83.239583
Batch [0199]/[0303]: acc-top1=61.523438 acc-top5=83.585938
Batch [0249]/[0303]: acc-top1=61.762500 acc-top5=83.825000
Batch [0299]/[0303]: acc-top1=61.479167 acc-top5=83.572917
[Epoch 042] training: accuracy=62.807098	 loss=1.534847
[Epoch 042] speed: 69 samples/sec	time cost: 3763.583081
[Epoch 042] validation: acc-top1=61.473804 acc-top5=83.565388 loss=1.671699
Epoch[043] Batch [0049]/[3759]	Speed: 48.362717 samples/sec	 accuracy=62.406250	 loss=1.544725	 lr=0.001000
Epoch[043] Batch [0099]/[3759]	Speed: 68.663799 samples/sec	 accuracy=61.859375	 loss=1.570092	 lr=0.001000
Epoch[043] Batch [0149]/[3759]	Speed: 69.713256 samples/sec	 accuracy=63.020833	 loss=1.530842	 lr=0.001000
Epoch[043] Batch [0199]/[3759]	Speed: 69.550791 samples/sec	 accuracy=62.921875	 loss=1.530100	 lr=0.001000
Epoch[043] Batch [0249]/[3759]	Speed: 66.701140 samples/sec	 accuracy=63.137500	 loss=1.517924	 lr=0.001000
Epoch[043] Batch [0299]/[3759]	Speed: 72.722937 samples/sec	 accuracy=62.937500	 loss=1.521373	 lr=0.001000
Epoch[043] Batch [0349]/[3759]	Speed: 70.141373 samples/sec	 accuracy=62.928571	 loss=1.522470	 lr=0.001000
Epoch[043] Batch [0399]/[3759]	Speed: 68.496645 samples/sec	 accuracy=62.867188	 loss=1.522983	 lr=0.001000
Epoch[043] Batch [0449]/[3759]	Speed: 70.261443 samples/sec	 accuracy=62.916667	 loss=1.524216	 lr=0.001000
Epoch[043] Batch [0499]/[3759]	Speed: 70.010114 samples/sec	 accuracy=63.109375	 loss=1.515934	 lr=0.001000
Epoch[043] Batch [0549]/[3759]	Speed: 68.691552 samples/sec	 accuracy=63.215909	 loss=1.511258	 lr=0.001000
Epoch[043] Batch [0599]/[3759]	Speed: 70.925142 samples/sec	 accuracy=63.190104	 loss=1.514627	 lr=0.001000
Epoch[043] Batch [0649]/[3759]	Speed: 69.111040 samples/sec	 accuracy=63.192308	 loss=1.514068	 lr=0.001000
Epoch[043] Batch [0699]/[3759]	Speed: 69.129897 samples/sec	 accuracy=63.212054	 loss=1.512735	 lr=0.001000
Epoch[043] Batch [0749]/[3759]	Speed: 71.010653 samples/sec	 accuracy=63.252083	 loss=1.508902	 lr=0.001000
Epoch[043] Batch [0799]/[3759]	Speed: 70.489908 samples/sec	 accuracy=63.220703	 loss=1.508976	 lr=0.001000
Epoch[043] Batch [0849]/[3759]	Speed: 69.006328 samples/sec	 accuracy=63.233456	 loss=1.509467	 lr=0.001000
Epoch[043] Batch [0899]/[3759]	Speed: 66.724579 samples/sec	 accuracy=63.217014	 loss=1.507205	 lr=0.001000
Epoch[043] Batch [0949]/[3759]	Speed: 71.419711 samples/sec	 accuracy=63.217105	 loss=1.507562	 lr=0.001000
Epoch[043] Batch [0999]/[3759]	Speed: 69.964629 samples/sec	 accuracy=63.229687	 loss=1.507523	 lr=0.001000
Epoch[043] Batch [1049]/[3759]	Speed: 68.775030 samples/sec	 accuracy=63.200893	 loss=1.509092	 lr=0.001000
Epoch[043] Batch [1099]/[3759]	Speed: 70.610376 samples/sec	 accuracy=63.220170	 loss=1.510043	 lr=0.001000
Epoch[043] Batch [1149]/[3759]	Speed: 70.177416 samples/sec	 accuracy=63.228261	 loss=1.510245	 lr=0.001000
Epoch[043] Batch [1199]/[3759]	Speed: 68.889781 samples/sec	 accuracy=63.253906	 loss=1.509230	 lr=0.001000
Epoch[043] Batch [1249]/[3759]	Speed: 69.107602 samples/sec	 accuracy=63.335000	 loss=1.506058	 lr=0.001000
Epoch[043] Batch [1299]/[3759]	Speed: 70.011374 samples/sec	 accuracy=63.366587	 loss=1.504173	 lr=0.001000
Epoch[043] Batch [1349]/[3759]	Speed: 69.169021 samples/sec	 accuracy=63.376157	 loss=1.504128	 lr=0.001000
Epoch[043] Batch [1399]/[3759]	Speed: 69.549636 samples/sec	 accuracy=63.422991	 loss=1.502598	 lr=0.001000
Epoch[043] Batch [1449]/[3759]	Speed: 69.521616 samples/sec	 accuracy=63.395474	 loss=1.504108	 lr=0.001000
Epoch[043] Batch [1499]/[3759]	Speed: 70.842152 samples/sec	 accuracy=63.382292	 loss=1.505029	 lr=0.001000
Epoch[043] Batch [1549]/[3759]	Speed: 70.635664 samples/sec	 accuracy=63.350806	 loss=1.506399	 lr=0.001000
Epoch[043] Batch [1599]/[3759]	Speed: 68.351696 samples/sec	 accuracy=63.314453	 loss=1.507518	 lr=0.001000
Epoch[043] Batch [1649]/[3759]	Speed: 69.238171 samples/sec	 accuracy=63.326705	 loss=1.507405	 lr=0.001000
Epoch[043] Batch [1699]/[3759]	Speed: 70.325517 samples/sec	 accuracy=63.319853	 loss=1.507503	 lr=0.001000
Epoch[043] Batch [1749]/[3759]	Speed: 70.731805 samples/sec	 accuracy=63.369643	 loss=1.505766	 lr=0.001000
Epoch[043] Batch [1799]/[3759]	Speed: 69.729581 samples/sec	 accuracy=63.315104	 loss=1.507364	 lr=0.001000
Epoch[043] Batch [1849]/[3759]	Speed: 69.621238 samples/sec	 accuracy=63.301520	 loss=1.508233	 lr=0.001000
Epoch[043] Batch [1899]/[3759]	Speed: 69.375938 samples/sec	 accuracy=63.301809	 loss=1.507981	 lr=0.001000
Epoch[043] Batch [1949]/[3759]	Speed: 68.569161 samples/sec	 accuracy=63.280449	 loss=1.509159	 lr=0.001000
Epoch[043] Batch [1999]/[3759]	Speed: 70.983478 samples/sec	 accuracy=63.274219	 loss=1.509353	 lr=0.001000
Epoch[043] Batch [2049]/[3759]	Speed: 68.889623 samples/sec	 accuracy=63.256860	 loss=1.509773	 lr=0.001000
Epoch[043] Batch [2099]/[3759]	Speed: 70.941450 samples/sec	 accuracy=63.206845	 loss=1.511937	 lr=0.001000
Epoch[043] Batch [2149]/[3759]	Speed: 69.758334 samples/sec	 accuracy=63.173692	 loss=1.512578	 lr=0.001000
Epoch[043] Batch [2199]/[3759]	Speed: 69.435736 samples/sec	 accuracy=63.154830	 loss=1.513772	 lr=0.001000
Epoch[043] Batch [2249]/[3759]	Speed: 71.058084 samples/sec	 accuracy=63.172222	 loss=1.512545	 lr=0.001000
Epoch[043] Batch [2299]/[3759]	Speed: 68.404506 samples/sec	 accuracy=63.173234	 loss=1.512256	 lr=0.001000
Epoch[043] Batch [2349]/[3759]	Speed: 70.931572 samples/sec	 accuracy=63.154920	 loss=1.512474	 lr=0.001000
Epoch[043] Batch [2399]/[3759]	Speed: 70.205221 samples/sec	 accuracy=63.138672	 loss=1.513196	 lr=0.001000
Epoch[043] Batch [2449]/[3759]	Speed: 67.867007 samples/sec	 accuracy=63.134566	 loss=1.512907	 lr=0.001000
Epoch[043] Batch [2499]/[3759]	Speed: 71.229179 samples/sec	 accuracy=63.163125	 loss=1.511933	 lr=0.001000
Epoch[043] Batch [2549]/[3759]	Speed: 69.124670 samples/sec	 accuracy=63.159314	 loss=1.511805	 lr=0.001000
Epoch[043] Batch [2599]/[3759]	Speed: 69.850217 samples/sec	 accuracy=63.168269	 loss=1.511285	 lr=0.001000
Epoch[043] Batch [2649]/[3759]	Speed: 68.577335 samples/sec	 accuracy=63.152123	 loss=1.511230	 lr=0.001000
Epoch[043] Batch [2699]/[3759]	Speed: 70.841000 samples/sec	 accuracy=63.164352	 loss=1.511291	 lr=0.001000
Epoch[043] Batch [2749]/[3759]	Speed: 70.078237 samples/sec	 accuracy=63.165341	 loss=1.511808	 lr=0.001000
Epoch[043] Batch [2799]/[3759]	Speed: 69.303213 samples/sec	 accuracy=63.169085	 loss=1.511790	 lr=0.001000
Epoch[043] Batch [2849]/[3759]	Speed: 70.628472 samples/sec	 accuracy=63.183114	 loss=1.511397	 lr=0.001000
Epoch[043] Batch [2899]/[3759]	Speed: 70.304959 samples/sec	 accuracy=63.194504	 loss=1.511285	 lr=0.001000
Epoch[043] Batch [2949]/[3759]	Speed: 69.221087 samples/sec	 accuracy=63.185381	 loss=1.511958	 lr=0.001000
Epoch[043] Batch [2999]/[3759]	Speed: 69.497281 samples/sec	 accuracy=63.180208	 loss=1.512265	 lr=0.001000
Epoch[043] Batch [3049]/[3759]	Speed: 68.659566 samples/sec	 accuracy=63.180328	 loss=1.512069	 lr=0.001000
Epoch[043] Batch [3099]/[3759]	Speed: 70.191987 samples/sec	 accuracy=63.172379	 loss=1.512621	 lr=0.001000
Epoch[043] Batch [3149]/[3759]	Speed: 70.478128 samples/sec	 accuracy=63.175595	 loss=1.512899	 lr=0.001000
Epoch[043] Batch [3199]/[3759]	Speed: 69.234846 samples/sec	 accuracy=63.198242	 loss=1.512349	 lr=0.001000
Epoch[043] Batch [3249]/[3759]	Speed: 70.094359 samples/sec	 accuracy=63.200962	 loss=1.512307	 lr=0.001000
Epoch[043] Batch [3299]/[3759]	Speed: 69.392778 samples/sec	 accuracy=63.196970	 loss=1.512383	 lr=0.001000
Epoch[043] Batch [3349]/[3759]	Speed: 70.699131 samples/sec	 accuracy=63.186567	 loss=1.512975	 lr=0.001000
Epoch[043] Batch [3399]/[3759]	Speed: 69.705011 samples/sec	 accuracy=63.171875	 loss=1.513554	 lr=0.001000
Epoch[043] Batch [3449]/[3759]	Speed: 68.491724 samples/sec	 accuracy=63.205163	 loss=1.512954	 lr=0.001000
Epoch[043] Batch [3499]/[3759]	Speed: 70.214780 samples/sec	 accuracy=63.197768	 loss=1.512672	 lr=0.001000
Epoch[043] Batch [3549]/[3759]	Speed: 69.155682 samples/sec	 accuracy=63.190141	 loss=1.513242	 lr=0.001000
Epoch[043] Batch [3599]/[3759]	Speed: 69.194238 samples/sec	 accuracy=63.200955	 loss=1.513009	 lr=0.001000
Epoch[043] Batch [3649]/[3759]	Speed: 70.889507 samples/sec	 accuracy=63.206336	 loss=1.513027	 lr=0.001000
Epoch[043] Batch [3699]/[3759]	Speed: 68.579337 samples/sec	 accuracy=63.208193	 loss=1.513273	 lr=0.001000
Epoch[043] Batch [3749]/[3759]	Speed: 78.160312 samples/sec	 accuracy=63.217917	 loss=1.512634	 lr=0.001000
Batch [0049]/[0303]: acc-top1=60.156250 acc-top5=82.875000
Batch [0099]/[0303]: acc-top1=60.390625 acc-top5=82.812500
Batch [0149]/[0303]: acc-top1=61.250000 acc-top5=83.260417
Batch [0199]/[0303]: acc-top1=61.742188 acc-top5=83.515625
Batch [0249]/[0303]: acc-top1=61.975000 acc-top5=83.712500
Batch [0299]/[0303]: acc-top1=61.723958 acc-top5=83.541667
[Epoch 043] training: accuracy=63.222433	 loss=1.512524
[Epoch 043] speed: 69 samples/sec	time cost: 3752.885546
[Epoch 043] validation: acc-top1=61.721328 acc-top5=83.544761 loss=1.666979
Epoch[044] Batch [0049]/[3760]	Speed: 47.112291 samples/sec	 accuracy=64.062500	 loss=1.492482	 lr=0.001000
Epoch[044] Batch [0099]/[3760]	Speed: 70.899409 samples/sec	 accuracy=64.265625	 loss=1.482471	 lr=0.001000
Epoch[044] Batch [0149]/[3760]	Speed: 69.898985 samples/sec	 accuracy=64.093750	 loss=1.488837	 lr=0.001000
Epoch[044] Batch [0199]/[3760]	Speed: 67.878632 samples/sec	 accuracy=63.796875	 loss=1.502022	 lr=0.001000
Epoch[044] Batch [0249]/[3760]	Speed: 71.403910 samples/sec	 accuracy=63.650000	 loss=1.509533	 lr=0.001000
Epoch[044] Batch [0299]/[3760]	Speed: 68.339691 samples/sec	 accuracy=63.661458	 loss=1.507032	 lr=0.001000
Epoch[044] Batch [0349]/[3760]	Speed: 69.996401 samples/sec	 accuracy=63.718750	 loss=1.504610	 lr=0.001000
Epoch[044] Batch [0399]/[3760]	Speed: 71.187463 samples/sec	 accuracy=63.761719	 loss=1.501318	 lr=0.001000
Epoch[044] Batch [0449]/[3760]	Speed: 69.572610 samples/sec	 accuracy=63.715278	 loss=1.497907	 lr=0.001000
Epoch[044] Batch [0499]/[3760]	Speed: 70.409474 samples/sec	 accuracy=63.712500	 loss=1.497013	 lr=0.001000
Epoch[044] Batch [0549]/[3760]	Speed: 68.749629 samples/sec	 accuracy=63.636364	 loss=1.503113	 lr=0.001000
Epoch[044] Batch [0599]/[3760]	Speed: 70.408088 samples/sec	 accuracy=63.664062	 loss=1.500862	 lr=0.001000
Epoch[044] Batch [0649]/[3760]	Speed: 68.407955 samples/sec	 accuracy=63.701923	 loss=1.500702	 lr=0.001000
Epoch[044] Batch [0699]/[3760]	Speed: 71.161158 samples/sec	 accuracy=63.636161	 loss=1.501035	 lr=0.001000
Epoch[044] Batch [0749]/[3760]	Speed: 70.118220 samples/sec	 accuracy=63.675000	 loss=1.500087	 lr=0.001000
Epoch[044] Batch [0799]/[3760]	Speed: 68.022875 samples/sec	 accuracy=63.693359	 loss=1.499234	 lr=0.001000
Epoch[044] Batch [0849]/[3760]	Speed: 70.857095 samples/sec	 accuracy=63.726103	 loss=1.498507	 lr=0.001000
Epoch[044] Batch [0899]/[3760]	Speed: 70.278795 samples/sec	 accuracy=63.727431	 loss=1.498565	 lr=0.001000
Epoch[044] Batch [0949]/[3760]	Speed: 67.834127 samples/sec	 accuracy=63.782895	 loss=1.498156	 lr=0.001000
Epoch[044] Batch [0999]/[3760]	Speed: 71.185389 samples/sec	 accuracy=63.878125	 loss=1.495294	 lr=0.001000
Epoch[044] Batch [1049]/[3760]	Speed: 69.368538 samples/sec	 accuracy=63.837798	 loss=1.496722	 lr=0.001000
Epoch[044] Batch [1099]/[3760]	Speed: 70.079300 samples/sec	 accuracy=63.829545	 loss=1.497585	 lr=0.001000
Epoch[044] Batch [1149]/[3760]	Speed: 69.991192 samples/sec	 accuracy=63.822011	 loss=1.500298	 lr=0.001000
Epoch[044] Batch [1199]/[3760]	Speed: 68.912891 samples/sec	 accuracy=63.781250	 loss=1.499659	 lr=0.001000
Epoch[044] Batch [1249]/[3760]	Speed: 68.164291 samples/sec	 accuracy=63.746250	 loss=1.501062	 lr=0.001000
Epoch[044] Batch [1299]/[3760]	Speed: 71.133874 samples/sec	 accuracy=63.760817	 loss=1.499531	 lr=0.001000
Epoch[044] Batch [1349]/[3760]	Speed: 69.128829 samples/sec	 accuracy=63.790509	 loss=1.498479	 lr=0.001000
Epoch[044] Batch [1399]/[3760]	Speed: 68.150755 samples/sec	 accuracy=63.781250	 loss=1.498193	 lr=0.001000
Epoch[044] Batch [1449]/[3760]	Speed: 70.715954 samples/sec	 accuracy=63.790948	 loss=1.497479	 lr=0.001000
Epoch[044] Batch [1499]/[3760]	Speed: 70.263711 samples/sec	 accuracy=63.819792	 loss=1.496219	 lr=0.001000
Epoch[044] Batch [1549]/[3760]	Speed: 69.408804 samples/sec	 accuracy=63.806452	 loss=1.497885	 lr=0.001000
Epoch[044] Batch [1599]/[3760]	Speed: 68.651783 samples/sec	 accuracy=63.826172	 loss=1.496253	 lr=0.001000
Epoch[044] Batch [1649]/[3760]	Speed: 68.597301 samples/sec	 accuracy=63.786932	 loss=1.497302	 lr=0.001000
Epoch[044] Batch [1699]/[3760]	Speed: 70.552415 samples/sec	 accuracy=63.816176	 loss=1.496384	 lr=0.001000
Epoch[044] Batch [1749]/[3760]	Speed: 70.113166 samples/sec	 accuracy=63.822321	 loss=1.495960	 lr=0.001000
Epoch[044] Batch [1799]/[3760]	Speed: 70.001478 samples/sec	 accuracy=63.803819	 loss=1.495698	 lr=0.001000
Epoch[044] Batch [1849]/[3760]	Speed: 67.974370 samples/sec	 accuracy=63.805743	 loss=1.495952	 lr=0.001000
Epoch[044] Batch [1899]/[3760]	Speed: 71.174277 samples/sec	 accuracy=63.782895	 loss=1.496079	 lr=0.001000
Epoch[044] Batch [1949]/[3760]	Speed: 69.119832 samples/sec	 accuracy=63.794872	 loss=1.496486	 lr=0.001000
Epoch[044] Batch [1999]/[3760]	Speed: 68.957627 samples/sec	 accuracy=63.795313	 loss=1.496304	 lr=0.001000
Epoch[044] Batch [2049]/[3760]	Speed: 69.605061 samples/sec	 accuracy=63.804116	 loss=1.495782	 lr=0.001000
Epoch[044] Batch [2099]/[3760]	Speed: 69.414855 samples/sec	 accuracy=63.816964	 loss=1.495817	 lr=0.001000
Epoch[044] Batch [2149]/[3760]	Speed: 69.429056 samples/sec	 accuracy=63.814680	 loss=1.495837	 lr=0.001000
Epoch[044] Batch [2199]/[3760]	Speed: 70.207360 samples/sec	 accuracy=63.826705	 loss=1.494593	 lr=0.001000
Epoch[044] Batch [2249]/[3760]	Speed: 67.153685 samples/sec	 accuracy=63.819444	 loss=1.494856	 lr=0.001000
Epoch[044] Batch [2299]/[3760]	Speed: 70.063304 samples/sec	 accuracy=63.819973	 loss=1.494678	 lr=0.001000
Epoch[044] Batch [2349]/[3760]	Speed: 69.692192 samples/sec	 accuracy=63.806516	 loss=1.494719	 lr=0.001000
Epoch[044] Batch [2399]/[3760]	Speed: 69.946012 samples/sec	 accuracy=63.807943	 loss=1.494281	 lr=0.001000
Epoch[044] Batch [2449]/[3760]	Speed: 70.019242 samples/sec	 accuracy=63.792092	 loss=1.493894	 lr=0.001000
Epoch[044] Batch [2499]/[3760]	Speed: 69.485491 samples/sec	 accuracy=63.793750	 loss=1.493792	 lr=0.001000
Epoch[044] Batch [2549]/[3760]	Speed: 69.518121 samples/sec	 accuracy=63.771446	 loss=1.494780	 lr=0.001000
Epoch[044] Batch [2599]/[3760]	Speed: 70.250874 samples/sec	 accuracy=63.762620	 loss=1.495045	 lr=0.001000
Epoch[044] Batch [2649]/[3760]	Speed: 70.975141 samples/sec	 accuracy=63.747052	 loss=1.494992	 lr=0.001000
Epoch[044] Batch [2699]/[3760]	Speed: 68.992130 samples/sec	 accuracy=63.730324	 loss=1.495898	 lr=0.001000
Epoch[044] Batch [2749]/[3760]	Speed: 68.686319 samples/sec	 accuracy=63.727841	 loss=1.495722	 lr=0.001000
Epoch[044] Batch [2799]/[3760]	Speed: 70.675735 samples/sec	 accuracy=63.733817	 loss=1.495758	 lr=0.001000
Epoch[044] Batch [2849]/[3760]	Speed: 70.932633 samples/sec	 accuracy=63.716009	 loss=1.496359	 lr=0.001000
Epoch[044] Batch [2899]/[3760]	Speed: 67.719046 samples/sec	 accuracy=63.724677	 loss=1.495978	 lr=0.001000
Epoch[044] Batch [2949]/[3760]	Speed: 68.978051 samples/sec	 accuracy=63.707097	 loss=1.497045	 lr=0.001000
Epoch[044] Batch [2999]/[3760]	Speed: 72.014789 samples/sec	 accuracy=63.711458	 loss=1.496982	 lr=0.001000
Epoch[044] Batch [3049]/[3760]	Speed: 67.712855 samples/sec	 accuracy=63.712602	 loss=1.496855	 lr=0.001000
Epoch[044] Batch [3099]/[3760]	Speed: 70.491724 samples/sec	 accuracy=63.689516	 loss=1.497452	 lr=0.001000
Epoch[044] Batch [3149]/[3760]	Speed: 69.865128 samples/sec	 accuracy=63.676091	 loss=1.497562	 lr=0.001000
Epoch[044] Batch [3199]/[3760]	Speed: 69.577318 samples/sec	 accuracy=63.669434	 loss=1.497590	 lr=0.001000
Epoch[044] Batch [3249]/[3760]	Speed: 69.648599 samples/sec	 accuracy=63.677885	 loss=1.497111	 lr=0.001000
Epoch[044] Batch [3299]/[3760]	Speed: 68.955921 samples/sec	 accuracy=63.675663	 loss=1.497328	 lr=0.001000
Epoch[044] Batch [3349]/[3760]	Speed: 68.133922 samples/sec	 accuracy=63.679104	 loss=1.497046	 lr=0.001000
Epoch[044] Batch [3399]/[3760]	Speed: 72.406511 samples/sec	 accuracy=63.676471	 loss=1.497240	 lr=0.001000
Epoch[044] Batch [3449]/[3760]	Speed: 67.369639 samples/sec	 accuracy=63.696558	 loss=1.496889	 lr=0.001000
Epoch[044] Batch [3499]/[3760]	Speed: 71.808141 samples/sec	 accuracy=63.704911	 loss=1.496751	 lr=0.001000
Epoch[044] Batch [3549]/[3760]	Speed: 69.346875 samples/sec	 accuracy=63.714349	 loss=1.496598	 lr=0.001000
Epoch[044] Batch [3599]/[3760]	Speed: 71.557645 samples/sec	 accuracy=63.720920	 loss=1.496339	 lr=0.001000
Epoch[044] Batch [3649]/[3760]	Speed: 68.335010 samples/sec	 accuracy=63.743151	 loss=1.495694	 lr=0.001000
Epoch[044] Batch [3699]/[3760]	Speed: 70.530393 samples/sec	 accuracy=63.733953	 loss=1.495676	 lr=0.001000
Epoch[044] Batch [3749]/[3760]	Speed: 73.141256 samples/sec	 accuracy=63.738750	 loss=1.495429	 lr=0.001000
Batch [0049]/[0303]: acc-top1=60.187500 acc-top5=82.906250
Batch [0099]/[0303]: acc-top1=60.593750 acc-top5=82.953125
Batch [0149]/[0303]: acc-top1=61.427083 acc-top5=83.458333
Batch [0199]/[0303]: acc-top1=61.859375 acc-top5=83.812500
Batch [0249]/[0303]: acc-top1=62.012500 acc-top5=84.006250
Batch [0299]/[0303]: acc-top1=61.729167 acc-top5=83.895833
[Epoch 044] training: accuracy=63.734624	 loss=1.495364
[Epoch 044] speed: 69 samples/sec	time cost: 3758.345999
[Epoch 044] validation: acc-top1=61.721328 acc-top5=83.900578 loss=1.648320
Epoch[045] Batch [0049]/[3760]	Speed: 48.645727 samples/sec	 accuracy=63.437500	 loss=1.476726	 lr=0.001000
Epoch[045] Batch [0099]/[3760]	Speed: 68.914822 samples/sec	 accuracy=64.843750	 loss=1.427018	 lr=0.001000
Epoch[045] Batch [0149]/[3760]	Speed: 69.168935 samples/sec	 accuracy=65.354167	 loss=1.423779	 lr=0.001000
Epoch[045] Batch [0199]/[3760]	Speed: 69.691880 samples/sec	 accuracy=64.914062	 loss=1.438736	 lr=0.001000
Epoch[045] Batch [0249]/[3760]	Speed: 69.494011 samples/sec	 accuracy=64.618750	 loss=1.455366	 lr=0.001000
Epoch[045] Batch [0299]/[3760]	Speed: 68.696279 samples/sec	 accuracy=64.770833	 loss=1.452811	 lr=0.001000
Epoch[045] Batch [0349]/[3760]	Speed: 69.766442 samples/sec	 accuracy=64.620536	 loss=1.457205	 lr=0.001000
Epoch[045] Batch [0399]/[3760]	Speed: 70.684013 samples/sec	 accuracy=64.726562	 loss=1.455859	 lr=0.001000
Epoch[045] Batch [0449]/[3760]	Speed: 69.776052 samples/sec	 accuracy=64.645833	 loss=1.457149	 lr=0.001000
Epoch[045] Batch [0499]/[3760]	Speed: 68.353285 samples/sec	 accuracy=64.593750	 loss=1.459555	 lr=0.001000
Epoch[045] Batch [0549]/[3760]	Speed: 69.814541 samples/sec	 accuracy=64.644886	 loss=1.457612	 lr=0.001000
Epoch[045] Batch [0599]/[3760]	Speed: 69.287616 samples/sec	 accuracy=64.567708	 loss=1.461065	 lr=0.001000
Epoch[045] Batch [0649]/[3760]	Speed: 68.527522 samples/sec	 accuracy=64.661058	 loss=1.460265	 lr=0.001000
Epoch[045] Batch [0699]/[3760]	Speed: 70.891119 samples/sec	 accuracy=64.683036	 loss=1.459563	 lr=0.001000
Epoch[045] Batch [0749]/[3760]	Speed: 71.051527 samples/sec	 accuracy=64.589583	 loss=1.463312	 lr=0.001000
Epoch[045] Batch [0799]/[3760]	Speed: 69.534508 samples/sec	 accuracy=64.564453	 loss=1.463823	 lr=0.001000
Epoch[045] Batch [0849]/[3760]	Speed: 68.745218 samples/sec	 accuracy=64.545956	 loss=1.463949	 lr=0.001000
Epoch[045] Batch [0899]/[3760]	Speed: 70.027505 samples/sec	 accuracy=64.439236	 loss=1.467398	 lr=0.001000
Epoch[045] Batch [0949]/[3760]	Speed: 69.633936 samples/sec	 accuracy=64.421053	 loss=1.466726	 lr=0.001000
Epoch[045] Batch [0999]/[3760]	Speed: 69.581079 samples/sec	 accuracy=64.390625	 loss=1.466617	 lr=0.001000
Epoch[045] Batch [1049]/[3760]	Speed: 66.887652 samples/sec	 accuracy=64.412202	 loss=1.465432	 lr=0.001000
Epoch[045] Batch [1099]/[3760]	Speed: 72.100378 samples/sec	 accuracy=64.367898	 loss=1.466942	 lr=0.001000
Epoch[045] Batch [1149]/[3760]	Speed: 70.016219 samples/sec	 accuracy=64.288043	 loss=1.469336	 lr=0.001000
Epoch[045] Batch [1199]/[3760]	Speed: 69.524165 samples/sec	 accuracy=64.230469	 loss=1.471877	 lr=0.001000
Epoch[045] Batch [1249]/[3760]	Speed: 69.390508 samples/sec	 accuracy=64.191250	 loss=1.473266	 lr=0.001000
Epoch[045] Batch [1299]/[3760]	Speed: 70.339604 samples/sec	 accuracy=64.181490	 loss=1.474943	 lr=0.001000
Epoch[045] Batch [1349]/[3760]	Speed: 69.930320 samples/sec	 accuracy=64.193287	 loss=1.474212	 lr=0.001000
Epoch[045] Batch [1399]/[3760]	Speed: 70.070485 samples/sec	 accuracy=64.169643	 loss=1.474728	 lr=0.001000
Epoch[045] Batch [1449]/[3760]	Speed: 67.669941 samples/sec	 accuracy=64.172414	 loss=1.474655	 lr=0.001000
Epoch[045] Batch [1499]/[3760]	Speed: 71.206883 samples/sec	 accuracy=64.188542	 loss=1.473783	 lr=0.001000
Epoch[045] Batch [1549]/[3760]	Speed: 68.928901 samples/sec	 accuracy=64.166331	 loss=1.474349	 lr=0.001000
Epoch[045] Batch [1599]/[3760]	Speed: 68.265163 samples/sec	 accuracy=64.161133	 loss=1.473567	 lr=0.001000
Epoch[045] Batch [1649]/[3760]	Speed: 70.953123 samples/sec	 accuracy=64.165720	 loss=1.473427	 lr=0.001000
Epoch[045] Batch [1699]/[3760]	Speed: 70.327286 samples/sec	 accuracy=64.162684	 loss=1.472442	 lr=0.001000
Epoch[045] Batch [1749]/[3760]	Speed: 69.328348 samples/sec	 accuracy=64.151786	 loss=1.473727	 lr=0.001000
Epoch[045] Batch [1799]/[3760]	Speed: 70.505940 samples/sec	 accuracy=64.157118	 loss=1.473737	 lr=0.001000
Epoch[045] Batch [1849]/[3760]	Speed: 69.001362 samples/sec	 accuracy=64.172297	 loss=1.473387	 lr=0.001000
Epoch[045] Batch [1899]/[3760]	Speed: 70.115211 samples/sec	 accuracy=64.168586	 loss=1.473230	 lr=0.001000
Epoch[045] Batch [1949]/[3760]	Speed: 68.901945 samples/sec	 accuracy=64.160256	 loss=1.473137	 lr=0.001000
Epoch[045] Batch [1999]/[3760]	Speed: 70.120862 samples/sec	 accuracy=64.185937	 loss=1.472211	 lr=0.001000
Epoch[045] Batch [2049]/[3760]	Speed: 69.470395 samples/sec	 accuracy=64.183689	 loss=1.471057	 lr=0.001000
Epoch[045] Batch [2099]/[3760]	Speed: 71.582038 samples/sec	 accuracy=64.157738	 loss=1.471999	 lr=0.001000
Epoch[045] Batch [2149]/[3760]	Speed: 65.298649 samples/sec	 accuracy=64.117733	 loss=1.473725	 lr=0.001000
Epoch[045] Batch [2199]/[3760]	Speed: 74.047565 samples/sec	 accuracy=64.096591	 loss=1.473656	 lr=0.001000
Epoch[045] Batch [2249]/[3760]	Speed: 69.439225 samples/sec	 accuracy=64.097917	 loss=1.473857	 lr=0.001000
Epoch[045] Batch [2299]/[3760]	Speed: 70.363327 samples/sec	 accuracy=64.111413	 loss=1.473198	 lr=0.001000
Epoch[045] Batch [2349]/[3760]	Speed: 69.972986 samples/sec	 accuracy=64.135638	 loss=1.473122	 lr=0.001000
Epoch[045] Batch [2399]/[3760]	Speed: 67.604902 samples/sec	 accuracy=64.136719	 loss=1.472605	 lr=0.001000
Epoch[045] Batch [2449]/[3760]	Speed: 70.911579 samples/sec	 accuracy=64.113520	 loss=1.472656	 lr=0.001000
Epoch[045] Batch [2499]/[3760]	Speed: 69.934637 samples/sec	 accuracy=64.113750	 loss=1.472798	 lr=0.001000
Epoch[045] Batch [2549]/[3760]	Speed: 70.511547 samples/sec	 accuracy=64.110907	 loss=1.472830	 lr=0.001000
Epoch[045] Batch [2599]/[3760]	Speed: 68.561758 samples/sec	 accuracy=64.115385	 loss=1.473171	 lr=0.001000
Epoch[045] Batch [2649]/[3760]	Speed: 70.473757 samples/sec	 accuracy=64.124410	 loss=1.472546	 lr=0.001000
Epoch[045] Batch [2699]/[3760]	Speed: 71.019233 samples/sec	 accuracy=64.109375	 loss=1.473419	 lr=0.001000
Epoch[045] Batch [2749]/[3760]	Speed: 69.846587 samples/sec	 accuracy=64.115909	 loss=1.473650	 lr=0.001000
Epoch[045] Batch [2799]/[3760]	Speed: 70.472597 samples/sec	 accuracy=64.096540	 loss=1.474607	 lr=0.001000
Epoch[045] Batch [2849]/[3760]	Speed: 68.582196 samples/sec	 accuracy=64.067982	 loss=1.475833	 lr=0.001000
Epoch[045] Batch [2899]/[3760]	Speed: 71.440341 samples/sec	 accuracy=64.077047	 loss=1.476291	 lr=0.001000
Epoch[045] Batch [2949]/[3760]	Speed: 69.659018 samples/sec	 accuracy=64.096928	 loss=1.475828	 lr=0.001000
Epoch[045] Batch [2999]/[3760]	Speed: 69.380238 samples/sec	 accuracy=64.097396	 loss=1.475123	 lr=0.001000
Epoch[045] Batch [3049]/[3760]	Speed: 69.799875 samples/sec	 accuracy=64.089139	 loss=1.475906	 lr=0.001000
Epoch[045] Batch [3099]/[3760]	Speed: 70.282455 samples/sec	 accuracy=64.100302	 loss=1.475243	 lr=0.001000
Epoch[045] Batch [3149]/[3760]	Speed: 70.279393 samples/sec	 accuracy=64.098710	 loss=1.475054	 lr=0.001000
Epoch[045] Batch [3199]/[3760]	Speed: 69.005580 samples/sec	 accuracy=64.092773	 loss=1.475201	 lr=0.001000
Epoch[045] Batch [3249]/[3760]	Speed: 69.997589 samples/sec	 accuracy=64.093269	 loss=1.475123	 lr=0.001000
Epoch[045] Batch [3299]/[3760]	Speed: 70.369675 samples/sec	 accuracy=64.091383	 loss=1.475461	 lr=0.001000
Epoch[045] Batch [3349]/[3760]	Speed: 69.156825 samples/sec	 accuracy=64.091884	 loss=1.475671	 lr=0.001000
Epoch[045] Batch [3399]/[3760]	Speed: 70.714929 samples/sec	 accuracy=64.090533	 loss=1.475280	 lr=0.001000
Epoch[045] Batch [3449]/[3760]	Speed: 68.818360 samples/sec	 accuracy=64.072464	 loss=1.476363	 lr=0.001000
Epoch[045] Batch [3499]/[3760]	Speed: 70.015556 samples/sec	 accuracy=64.100446	 loss=1.475212	 lr=0.001000
Epoch[045] Batch [3549]/[3760]	Speed: 69.929353 samples/sec	 accuracy=64.113556	 loss=1.474663	 lr=0.001000
Epoch[045] Batch [3599]/[3760]	Speed: 69.088420 samples/sec	 accuracy=64.122396	 loss=1.474905	 lr=0.001000
Epoch[045] Batch [3649]/[3760]	Speed: 70.127399 samples/sec	 accuracy=64.118579	 loss=1.474804	 lr=0.001000
Epoch[045] Batch [3699]/[3760]	Speed: 67.438492 samples/sec	 accuracy=64.127111	 loss=1.474650	 lr=0.001000
Epoch[045] Batch [3749]/[3760]	Speed: 78.924020 samples/sec	 accuracy=64.124583	 loss=1.474749	 lr=0.001000
Batch [0049]/[0303]: acc-top1=61.281250 acc-top5=83.343750
Batch [0099]/[0303]: acc-top1=60.984375 acc-top5=83.156250
Batch [0149]/[0303]: acc-top1=61.833333 acc-top5=83.458333
Batch [0199]/[0303]: acc-top1=62.187500 acc-top5=83.804688
Batch [0249]/[0303]: acc-top1=62.393750 acc-top5=83.962500
Batch [0299]/[0303]: acc-top1=62.088542 acc-top5=83.833333
[Epoch 045] training: accuracy=64.131067	 loss=1.474489
[Epoch 045] speed: 69 samples/sec	time cost: 3751.599510
[Epoch 045] validation: acc-top1=62.092616 acc-top5=83.864480 loss=1.647693
Epoch[046] Batch [0049]/[3759]	Speed: 48.663599 samples/sec	 accuracy=64.687500	 loss=1.448877	 lr=0.001000
Epoch[046] Batch [0099]/[3759]	Speed: 68.434215 samples/sec	 accuracy=64.312500	 loss=1.457801	 lr=0.001000
Epoch[046] Batch [0149]/[3759]	Speed: 69.676251 samples/sec	 accuracy=64.447917	 loss=1.451724	 lr=0.001000
Epoch[046] Batch [0199]/[3759]	Speed: 70.360136 samples/sec	 accuracy=64.460938	 loss=1.462193	 lr=0.001000
Epoch[046] Batch [0249]/[3759]	Speed: 71.242267 samples/sec	 accuracy=64.281250	 loss=1.463788	 lr=0.001000
Epoch[046] Batch [0299]/[3759]	Speed: 68.807243 samples/sec	 accuracy=64.125000	 loss=1.468735	 lr=0.001000
Epoch[046] Batch [0349]/[3759]	Speed: 69.919040 samples/sec	 accuracy=64.040179	 loss=1.472130	 lr=0.001000
Epoch[046] Batch [0399]/[3759]	Speed: 69.182063 samples/sec	 accuracy=64.027344	 loss=1.471006	 lr=0.001000
Epoch[046] Batch [0449]/[3759]	Speed: 70.194287 samples/sec	 accuracy=64.041667	 loss=1.473828	 lr=0.001000
Epoch[046] Batch [0499]/[3759]	Speed: 69.482518 samples/sec	 accuracy=63.950000	 loss=1.477164	 lr=0.001000
Epoch[046] Batch [0549]/[3759]	Speed: 70.285358 samples/sec	 accuracy=64.005682	 loss=1.472233	 lr=0.001000
Epoch[046] Batch [0599]/[3759]	Speed: 69.136258 samples/sec	 accuracy=63.976562	 loss=1.473923	 lr=0.001000
Epoch[046] Batch [0649]/[3759]	Speed: 69.452440 samples/sec	 accuracy=63.903846	 loss=1.475152	 lr=0.001000
Epoch[046] Batch [0699]/[3759]	Speed: 68.990106 samples/sec	 accuracy=63.939732	 loss=1.472606	 lr=0.001000
Epoch[046] Batch [0749]/[3759]	Speed: 69.905327 samples/sec	 accuracy=63.964583	 loss=1.471221	 lr=0.001000
Epoch[046] Batch [0799]/[3759]	Speed: 70.374497 samples/sec	 accuracy=64.019531	 loss=1.470428	 lr=0.001000
Epoch[046] Batch [0849]/[3759]	Speed: 69.673913 samples/sec	 accuracy=64.040441	 loss=1.471892	 lr=0.001000
Epoch[046] Batch [0899]/[3759]	Speed: 69.875670 samples/sec	 accuracy=64.142361	 loss=1.467750	 lr=0.001000
Epoch[046] Batch [0949]/[3759]	Speed: 69.543530 samples/sec	 accuracy=64.129934	 loss=1.468373	 lr=0.001000
Epoch[046] Batch [0999]/[3759]	Speed: 70.532042 samples/sec	 accuracy=64.195312	 loss=1.463941	 lr=0.001000
Epoch[046] Batch [1049]/[3759]	Speed: 68.377968 samples/sec	 accuracy=64.110119	 loss=1.465557	 lr=0.001000
Epoch[046] Batch [1099]/[3759]	Speed: 68.896294 samples/sec	 accuracy=64.132102	 loss=1.464123	 lr=0.001000
Epoch[046] Batch [1149]/[3759]	Speed: 72.005255 samples/sec	 accuracy=64.150815	 loss=1.464809	 lr=0.001000
Epoch[046] Batch [1199]/[3759]	Speed: 68.024524 samples/sec	 accuracy=64.104167	 loss=1.466880	 lr=0.001000
Epoch[046] Batch [1249]/[3759]	Speed: 70.213724 samples/sec	 accuracy=64.045000	 loss=1.468454	 lr=0.001000
Epoch[046] Batch [1299]/[3759]	Speed: 69.530486 samples/sec	 accuracy=64.063702	 loss=1.466349	 lr=0.001000
Epoch[046] Batch [1349]/[3759]	Speed: 71.243483 samples/sec	 accuracy=64.101852	 loss=1.464464	 lr=0.001000
Epoch[046] Batch [1399]/[3759]	Speed: 70.002260 samples/sec	 accuracy=64.117188	 loss=1.465083	 lr=0.001000
Epoch[046] Batch [1449]/[3759]	Speed: 69.598827 samples/sec	 accuracy=64.156250	 loss=1.465107	 lr=0.001000
Epoch[046] Batch [1499]/[3759]	Speed: 67.815361 samples/sec	 accuracy=64.162500	 loss=1.466053	 lr=0.001000
Epoch[046] Batch [1549]/[3759]	Speed: 70.255368 samples/sec	 accuracy=64.221774	 loss=1.464963	 lr=0.001000
Epoch[046] Batch [1599]/[3759]	Speed: 71.531564 samples/sec	 accuracy=64.197266	 loss=1.466701	 lr=0.001000
Epoch[046] Batch [1649]/[3759]	Speed: 68.268329 samples/sec	 accuracy=64.189394	 loss=1.467673	 lr=0.001000
Epoch[046] Batch [1699]/[3759]	Speed: 69.989229 samples/sec	 accuracy=64.217831	 loss=1.467028	 lr=0.001000
Epoch[046] Batch [1749]/[3759]	Speed: 68.770641 samples/sec	 accuracy=64.216071	 loss=1.465889	 lr=0.001000
Epoch[046] Batch [1799]/[3759]	Speed: 70.817127 samples/sec	 accuracy=64.195312	 loss=1.466105	 lr=0.001000
Epoch[046] Batch [1849]/[3759]	Speed: 70.214228 samples/sec	 accuracy=64.195946	 loss=1.466257	 lr=0.001000
Epoch[046] Batch [1899]/[3759]	Speed: 70.088488 samples/sec	 accuracy=64.208059	 loss=1.465919	 lr=0.001000
Epoch[046] Batch [1949]/[3759]	Speed: 66.834269 samples/sec	 accuracy=64.227564	 loss=1.466000	 lr=0.001000
Epoch[046] Batch [1999]/[3759]	Speed: 71.338259 samples/sec	 accuracy=64.221094	 loss=1.466125	 lr=0.001000
Epoch[046] Batch [2049]/[3759]	Speed: 70.785477 samples/sec	 accuracy=64.239329	 loss=1.464815	 lr=0.001000
Epoch[046] Batch [2099]/[3759]	Speed: 69.294028 samples/sec	 accuracy=64.219494	 loss=1.464599	 lr=0.001000
Epoch[046] Batch [2149]/[3759]	Speed: 69.267733 samples/sec	 accuracy=64.230378	 loss=1.465497	 lr=0.001000
Epoch[046] Batch [2199]/[3759]	Speed: 70.442534 samples/sec	 accuracy=64.227983	 loss=1.465974	 lr=0.001000
Epoch[046] Batch [2249]/[3759]	Speed: 69.007570 samples/sec	 accuracy=64.206250	 loss=1.465959	 lr=0.001000
Epoch[046] Batch [2299]/[3759]	Speed: 70.174497 samples/sec	 accuracy=64.243207	 loss=1.464656	 lr=0.001000
Epoch[046] Batch [2349]/[3759]	Speed: 70.274181 samples/sec	 accuracy=64.261968	 loss=1.463739	 lr=0.001000
Epoch[046] Batch [2399]/[3759]	Speed: 69.137525 samples/sec	 accuracy=64.274089	 loss=1.464061	 lr=0.001000
Epoch[046] Batch [2449]/[3759]	Speed: 69.687589 samples/sec	 accuracy=64.258291	 loss=1.464217	 lr=0.001000
Epoch[046] Batch [2499]/[3759]	Speed: 71.019820 samples/sec	 accuracy=64.265625	 loss=1.464417	 lr=0.001000
Epoch[046] Batch [2549]/[3759]	Speed: 68.425835 samples/sec	 accuracy=64.265931	 loss=1.464140	 lr=0.001000
Epoch[046] Batch [2599]/[3759]	Speed: 69.612045 samples/sec	 accuracy=64.234976	 loss=1.464872	 lr=0.001000
Epoch[046] Batch [2649]/[3759]	Speed: 70.608527 samples/sec	 accuracy=64.266509	 loss=1.464173	 lr=0.001000
Epoch[046] Batch [2699]/[3759]	Speed: 69.281665 samples/sec	 accuracy=64.281829	 loss=1.463528	 lr=0.001000
Epoch[046] Batch [2749]/[3759]	Speed: 69.485023 samples/sec	 accuracy=64.292045	 loss=1.463423	 lr=0.001000
Epoch[046] Batch [2799]/[3759]	Speed: 69.744744 samples/sec	 accuracy=64.304129	 loss=1.463391	 lr=0.001000
Epoch[046] Batch [2849]/[3759]	Speed: 69.336253 samples/sec	 accuracy=64.290570	 loss=1.463749	 lr=0.001000
Epoch[046] Batch [2899]/[3759]	Speed: 70.805690 samples/sec	 accuracy=64.296336	 loss=1.463698	 lr=0.001000
Epoch[046] Batch [2949]/[3759]	Speed: 69.032787 samples/sec	 accuracy=64.293962	 loss=1.463414	 lr=0.001000
Epoch[046] Batch [2999]/[3759]	Speed: 70.002883 samples/sec	 accuracy=64.284896	 loss=1.463973	 lr=0.001000
Epoch[046] Batch [3049]/[3759]	Speed: 69.395634 samples/sec	 accuracy=64.300205	 loss=1.463847	 lr=0.001000
Epoch[046] Batch [3099]/[3759]	Speed: 69.867589 samples/sec	 accuracy=64.299899	 loss=1.463863	 lr=0.001000
Epoch[046] Batch [3149]/[3759]	Speed: 69.908213 samples/sec	 accuracy=64.289187	 loss=1.464051	 lr=0.001000
Epoch[046] Batch [3199]/[3759]	Speed: 68.677306 samples/sec	 accuracy=64.310547	 loss=1.463113	 lr=0.001000
Epoch[046] Batch [3249]/[3759]	Speed: 70.377043 samples/sec	 accuracy=64.310577	 loss=1.463659	 lr=0.001000
Epoch[046] Batch [3299]/[3759]	Speed: 69.022589 samples/sec	 accuracy=64.290720	 loss=1.463967	 lr=0.001000
Epoch[046] Batch [3349]/[3759]	Speed: 70.476880 samples/sec	 accuracy=64.296642	 loss=1.463953	 lr=0.001000
Epoch[046] Batch [3399]/[3759]	Speed: 69.067279 samples/sec	 accuracy=64.312500	 loss=1.463772	 lr=0.001000
Epoch[046] Batch [3449]/[3759]	Speed: 70.309442 samples/sec	 accuracy=64.298913	 loss=1.464074	 lr=0.001000
Epoch[046] Batch [3499]/[3759]	Speed: 70.558081 samples/sec	 accuracy=64.289286	 loss=1.464505	 lr=0.001000
Epoch[046] Batch [3549]/[3759]	Speed: 70.803336 samples/sec	 accuracy=64.278169	 loss=1.465216	 lr=0.001000
Epoch[046] Batch [3599]/[3759]	Speed: 69.233878 samples/sec	 accuracy=64.272569	 loss=1.465731	 lr=0.001000
Epoch[046] Batch [3649]/[3759]	Speed: 69.462454 samples/sec	 accuracy=64.269264	 loss=1.466184	 lr=0.001000
Epoch[046] Batch [3699]/[3759]	Speed: 69.770937 samples/sec	 accuracy=64.269426	 loss=1.466319	 lr=0.001000
Epoch[046] Batch [3749]/[3759]	Speed: 76.687500 samples/sec	 accuracy=64.252083	 loss=1.466405	 lr=0.001000
Batch [0049]/[0303]: acc-top1=60.687500 acc-top5=83.000000
Batch [0099]/[0303]: acc-top1=60.750000 acc-top5=83.000000
Batch [0149]/[0303]: acc-top1=61.645833 acc-top5=83.479167
Batch [0199]/[0303]: acc-top1=61.992187 acc-top5=83.867188
Batch [0249]/[0303]: acc-top1=62.156250 acc-top5=84.093750
Batch [0299]/[0303]: acc-top1=61.921875 acc-top5=83.942708
[Epoch 046] training: accuracy=64.249967	 loss=1.466337
[Epoch 046] speed: 69 samples/sec	time cost: 3752.771460
[Epoch 046] validation: acc-top1=61.937913 acc-top5=83.952145 loss=1.642159
Epoch[047] Batch [0049]/[3760]	Speed: 48.642066 samples/sec	 accuracy=64.156250	 loss=1.454132	 lr=0.001000
Epoch[047] Batch [0099]/[3760]	Speed: 67.498882 samples/sec	 accuracy=65.046875	 loss=1.431774	 lr=0.001000
Epoch[047] Batch [0149]/[3760]	Speed: 70.142767 samples/sec	 accuracy=65.072917	 loss=1.430676	 lr=0.001000
Epoch[047] Batch [0199]/[3760]	Speed: 69.753713 samples/sec	 accuracy=65.187500	 loss=1.426908	 lr=0.001000
Epoch[047] Batch [0249]/[3760]	Speed: 69.904739 samples/sec	 accuracy=65.206250	 loss=1.424924	 lr=0.001000
Epoch[047] Batch [0299]/[3760]	Speed: 68.223920 samples/sec	 accuracy=65.140625	 loss=1.426077	 lr=0.001000
Epoch[047] Batch [0349]/[3760]	Speed: 70.571997 samples/sec	 accuracy=65.138393	 loss=1.428463	 lr=0.001000
Epoch[047] Batch [0399]/[3760]	Speed: 70.329421 samples/sec	 accuracy=64.988281	 loss=1.432950	 lr=0.001000
Epoch[047] Batch [0449]/[3760]	Speed: 69.371459 samples/sec	 accuracy=65.107639	 loss=1.429052	 lr=0.001000
Epoch[047] Batch [0499]/[3760]	Speed: 70.063240 samples/sec	 accuracy=64.993750	 loss=1.431230	 lr=0.001000
Epoch[047] Batch [0549]/[3760]	Speed: 70.023780 samples/sec	 accuracy=64.985795	 loss=1.435160	 lr=0.001000
Epoch[047] Batch [0599]/[3760]	Speed: 70.658399 samples/sec	 accuracy=65.018229	 loss=1.435488	 lr=0.001000
Epoch[047] Batch [0649]/[3760]	Speed: 68.225388 samples/sec	 accuracy=64.932692	 loss=1.436560	 lr=0.001000
Epoch[047] Batch [0699]/[3760]	Speed: 70.117376 samples/sec	 accuracy=64.901786	 loss=1.439427	 lr=0.001000
Epoch[047] Batch [0749]/[3760]	Speed: 69.274035 samples/sec	 accuracy=64.897917	 loss=1.438477	 lr=0.001000
Epoch[047] Batch [0799]/[3760]	Speed: 70.859724 samples/sec	 accuracy=64.916016	 loss=1.435560	 lr=0.001000
Epoch[047] Batch [0849]/[3760]	Speed: 69.225099 samples/sec	 accuracy=64.909926	 loss=1.436430	 lr=0.001000
Epoch[047] Batch [0899]/[3760]	Speed: 69.450926 samples/sec	 accuracy=64.822917	 loss=1.437827	 lr=0.001000
Epoch[047] Batch [0949]/[3760]	Speed: 70.764313 samples/sec	 accuracy=64.768092	 loss=1.439003	 lr=0.001000
Epoch[047] Batch [0999]/[3760]	Speed: 69.857241 samples/sec	 accuracy=64.726562	 loss=1.442040	 lr=0.001000
Epoch[047] Batch [1049]/[3760]	Speed: 70.510431 samples/sec	 accuracy=64.706845	 loss=1.442774	 lr=0.001000
Epoch[047] Batch [1099]/[3760]	Speed: 68.139260 samples/sec	 accuracy=64.741477	 loss=1.442202	 lr=0.001000
Epoch[047] Batch [1149]/[3760]	Speed: 71.062845 samples/sec	 accuracy=64.698370	 loss=1.445415	 lr=0.001000
Epoch[047] Batch [1199]/[3760]	Speed: 69.466562 samples/sec	 accuracy=64.617188	 loss=1.447291	 lr=0.001000
Epoch[047] Batch [1249]/[3760]	Speed: 69.251713 samples/sec	 accuracy=64.592500	 loss=1.446599	 lr=0.001000
Epoch[047] Batch [1299]/[3760]	Speed: 69.172026 samples/sec	 accuracy=64.588942	 loss=1.447525	 lr=0.001000
Epoch[047] Batch [1349]/[3760]	Speed: 70.816244 samples/sec	 accuracy=64.594907	 loss=1.447355	 lr=0.001000
Epoch[047] Batch [1399]/[3760]	Speed: 69.474924 samples/sec	 accuracy=64.561384	 loss=1.447702	 lr=0.001000
Epoch[047] Batch [1449]/[3760]	Speed: 69.236187 samples/sec	 accuracy=64.587284	 loss=1.446550	 lr=0.001000
Epoch[047] Batch [1499]/[3760]	Speed: 69.423293 samples/sec	 accuracy=64.586458	 loss=1.447775	 lr=0.001000
Epoch[047] Batch [1549]/[3760]	Speed: 69.959582 samples/sec	 accuracy=64.570565	 loss=1.448623	 lr=0.001000
Epoch[047] Batch [1599]/[3760]	Speed: 70.140606 samples/sec	 accuracy=64.530273	 loss=1.449790	 lr=0.001000
Epoch[047] Batch [1649]/[3760]	Speed: 69.633796 samples/sec	 accuracy=64.552083	 loss=1.449994	 lr=0.001000
Epoch[047] Batch [1699]/[3760]	Speed: 70.064937 samples/sec	 accuracy=64.515625	 loss=1.451477	 lr=0.001000
Epoch[047] Batch [1749]/[3760]	Speed: 70.542611 samples/sec	 accuracy=64.501786	 loss=1.451806	 lr=0.001000
Epoch[047] Batch [1799]/[3760]	Speed: 69.671264 samples/sec	 accuracy=64.484375	 loss=1.452624	 lr=0.001000
Epoch[047] Batch [1849]/[3760]	Speed: 69.398277 samples/sec	 accuracy=64.478041	 loss=1.452993	 lr=0.001000
Epoch[047] Batch [1899]/[3760]	Speed: 69.847362 samples/sec	 accuracy=64.453947	 loss=1.453519	 lr=0.001000
Epoch[047] Batch [1949]/[3760]	Speed: 69.306967 samples/sec	 accuracy=64.419872	 loss=1.454711	 lr=0.001000
Epoch[047] Batch [1999]/[3760]	Speed: 70.528113 samples/sec	 accuracy=64.448437	 loss=1.453641	 lr=0.001000
Epoch[047] Batch [2049]/[3760]	Speed: 69.550691 samples/sec	 accuracy=64.489329	 loss=1.453280	 lr=0.001000
Epoch[047] Batch [2099]/[3760]	Speed: 69.291411 samples/sec	 accuracy=64.457589	 loss=1.454449	 lr=0.001000
Epoch[047] Batch [2149]/[3760]	Speed: 68.423208 samples/sec	 accuracy=64.455669	 loss=1.454420	 lr=0.001000
Epoch[047] Batch [2199]/[3760]	Speed: 71.217919 samples/sec	 accuracy=64.446733	 loss=1.455329	 lr=0.001000
Epoch[047] Batch [2249]/[3760]	Speed: 70.690435 samples/sec	 accuracy=64.461111	 loss=1.454842	 lr=0.001000
Epoch[047] Batch [2299]/[3760]	Speed: 69.302407 samples/sec	 accuracy=64.468071	 loss=1.454915	 lr=0.001000
Epoch[047] Batch [2349]/[3760]	Speed: 71.277454 samples/sec	 accuracy=64.483378	 loss=1.455804	 lr=0.001000
Epoch[047] Batch [2399]/[3760]	Speed: 68.997709 samples/sec	 accuracy=64.497396	 loss=1.455095	 lr=0.001000
Epoch[047] Batch [2449]/[3760]	Speed: 69.241355 samples/sec	 accuracy=64.498087	 loss=1.455651	 lr=0.001000
Epoch[047] Batch [2499]/[3760]	Speed: 70.033015 samples/sec	 accuracy=64.493125	 loss=1.455661	 lr=0.001000
Epoch[047] Batch [2549]/[3760]	Speed: 69.816126 samples/sec	 accuracy=64.488971	 loss=1.456102	 lr=0.001000
Epoch[047] Batch [2599]/[3760]	Speed: 67.210013 samples/sec	 accuracy=64.481370	 loss=1.456686	 lr=0.001000
Epoch[047] Batch [2649]/[3760]	Speed: 72.410632 samples/sec	 accuracy=64.500000	 loss=1.456114	 lr=0.001000
Epoch[047] Batch [2699]/[3760]	Speed: 70.435549 samples/sec	 accuracy=64.526620	 loss=1.454992	 lr=0.001000
Epoch[047] Batch [2749]/[3760]	Speed: 68.907777 samples/sec	 accuracy=64.515909	 loss=1.455485	 lr=0.001000
Epoch[047] Batch [2799]/[3760]	Speed: 70.789378 samples/sec	 accuracy=64.503906	 loss=1.456486	 lr=0.001000
Epoch[047] Batch [2849]/[3760]	Speed: 67.340880 samples/sec	 accuracy=64.503289	 loss=1.456595	 lr=0.001000
Epoch[047] Batch [2899]/[3760]	Speed: 71.873879 samples/sec	 accuracy=64.497845	 loss=1.456997	 lr=0.001000
Epoch[047] Batch [2949]/[3760]	Speed: 70.216818 samples/sec	 accuracy=64.502119	 loss=1.456948	 lr=0.001000
Epoch[047] Batch [2999]/[3760]	Speed: 68.313177 samples/sec	 accuracy=64.507292	 loss=1.457070	 lr=0.001000
Epoch[047] Batch [3049]/[3760]	Speed: 70.775252 samples/sec	 accuracy=64.507172	 loss=1.457118	 lr=0.001000
Epoch[047] Batch [3099]/[3760]	Speed: 70.722225 samples/sec	 accuracy=64.502016	 loss=1.457436	 lr=0.001000
Epoch[047] Batch [3149]/[3760]	Speed: 66.005034 samples/sec	 accuracy=64.501488	 loss=1.457372	 lr=0.001000
Epoch[047] Batch [3199]/[3760]	Speed: 72.924175 samples/sec	 accuracy=64.506836	 loss=1.457460	 lr=0.001000
Epoch[047] Batch [3249]/[3760]	Speed: 67.602971 samples/sec	 accuracy=64.506731	 loss=1.457658	 lr=0.001000
Epoch[047] Batch [3299]/[3760]	Speed: 71.427026 samples/sec	 accuracy=64.515152	 loss=1.457295	 lr=0.001000
Epoch[047] Batch [3349]/[3760]	Speed: 70.123803 samples/sec	 accuracy=64.502332	 loss=1.457643	 lr=0.001000
Epoch[047] Batch [3399]/[3760]	Speed: 69.887180 samples/sec	 accuracy=64.522978	 loss=1.457905	 lr=0.001000
Epoch[047] Batch [3449]/[3760]	Speed: 70.178884 samples/sec	 accuracy=64.551178	 loss=1.456607	 lr=0.001000
Epoch[047] Batch [3499]/[3760]	Speed: 68.401278 samples/sec	 accuracy=64.549554	 loss=1.456914	 lr=0.001000
Epoch[047] Batch [3549]/[3760]	Speed: 71.231270 samples/sec	 accuracy=64.551056	 loss=1.456738	 lr=0.001000
Epoch[047] Batch [3599]/[3760]	Speed: 68.792711 samples/sec	 accuracy=64.561198	 loss=1.456727	 lr=0.001000
Epoch[047] Batch [3649]/[3760]	Speed: 70.794466 samples/sec	 accuracy=64.562500	 loss=1.456853	 lr=0.001000
Epoch[047] Batch [3699]/[3760]	Speed: 69.349263 samples/sec	 accuracy=64.552787	 loss=1.456481	 lr=0.001000
Epoch[047] Batch [3749]/[3760]	Speed: 77.324292 samples/sec	 accuracy=64.545417	 loss=1.456550	 lr=0.001000
Batch [0049]/[0303]: acc-top1=60.937500 acc-top5=83.750000
Batch [0099]/[0303]: acc-top1=61.093750 acc-top5=83.437500
Batch [0149]/[0303]: acc-top1=61.916667 acc-top5=83.781250
Batch [0199]/[0303]: acc-top1=62.453125 acc-top5=83.953125
Batch [0249]/[0303]: acc-top1=62.625000 acc-top5=84.175000
Batch [0299]/[0303]: acc-top1=62.359375 acc-top5=83.921875
[Epoch 047] training: accuracy=64.540392	 loss=1.456678
[Epoch 047] speed: 69 samples/sec	time cost: 3750.830847
[Epoch 047] validation: acc-top1=62.360767 acc-top5=83.910891 loss=1.639234
Epoch[048] Batch [0049]/[3760]	Speed: 47.856872 samples/sec	 accuracy=64.312500	 loss=1.441969	 lr=0.001000
Epoch[048] Batch [0099]/[3760]	Speed: 68.109626 samples/sec	 accuracy=65.234375	 loss=1.421236	 lr=0.001000
Epoch[048] Batch [0149]/[3760]	Speed: 68.127750 samples/sec	 accuracy=64.916667	 loss=1.440039	 lr=0.001000
Epoch[048] Batch [0199]/[3760]	Speed: 71.430920 samples/sec	 accuracy=64.992188	 loss=1.438810	 lr=0.001000
Epoch[048] Batch [0249]/[3760]	Speed: 70.523225 samples/sec	 accuracy=64.900000	 loss=1.440132	 lr=0.001000
Epoch[048] Batch [0299]/[3760]	Speed: 68.215262 samples/sec	 accuracy=64.734375	 loss=1.447171	 lr=0.001000
Epoch[048] Batch [0349]/[3760]	Speed: 70.844486 samples/sec	 accuracy=64.513393	 loss=1.448556	 lr=0.001000
Epoch[048] Batch [0399]/[3760]	Speed: 69.255056 samples/sec	 accuracy=64.675781	 loss=1.447346	 lr=0.001000
Epoch[048] Batch [0449]/[3760]	Speed: 69.793901 samples/sec	 accuracy=64.753472	 loss=1.444930	 lr=0.001000
Epoch[048] Batch [0499]/[3760]	Speed: 69.266334 samples/sec	 accuracy=64.712500	 loss=1.445127	 lr=0.001000
Epoch[048] Batch [0549]/[3760]	Speed: 70.104516 samples/sec	 accuracy=64.724432	 loss=1.446474	 lr=0.001000
Epoch[048] Batch [0599]/[3760]	Speed: 68.361345 samples/sec	 accuracy=64.682292	 loss=1.449176	 lr=0.001000
Epoch[048] Batch [0649]/[3760]	Speed: 70.949805 samples/sec	 accuracy=64.663462	 loss=1.450067	 lr=0.001000
Epoch[048] Batch [0699]/[3760]	Speed: 70.205879 samples/sec	 accuracy=64.622768	 loss=1.452144	 lr=0.001000
Epoch[048] Batch [0749]/[3760]	Speed: 70.218102 samples/sec	 accuracy=64.643750	 loss=1.450564	 lr=0.001000
Epoch[048] Batch [0799]/[3760]	Speed: 69.510304 samples/sec	 accuracy=64.664062	 loss=1.450705	 lr=0.001000
Epoch[048] Batch [0849]/[3760]	Speed: 68.652915 samples/sec	 accuracy=64.678309	 loss=1.450072	 lr=0.001000
Epoch[048] Batch [0899]/[3760]	Speed: 69.670227 samples/sec	 accuracy=64.696181	 loss=1.450792	 lr=0.001000
Epoch[048] Batch [0949]/[3760]	Speed: 69.628398 samples/sec	 accuracy=64.736842	 loss=1.449266	 lr=0.001000
Epoch[048] Batch [0999]/[3760]	Speed: 70.233244 samples/sec	 accuracy=64.692188	 loss=1.451926	 lr=0.001000
Epoch[048] Batch [1049]/[3760]	Speed: 70.090890 samples/sec	 accuracy=64.681548	 loss=1.452091	 lr=0.001000
Epoch[048] Batch [1099]/[3760]	Speed: 67.585278 samples/sec	 accuracy=64.661932	 loss=1.452571	 lr=0.001000
Epoch[048] Batch [1149]/[3760]	Speed: 71.847311 samples/sec	 accuracy=64.657609	 loss=1.451611	 lr=0.001000
Epoch[048] Batch [1199]/[3760]	Speed: 68.868136 samples/sec	 accuracy=64.656250	 loss=1.451388	 lr=0.001000
Epoch[048] Batch [1249]/[3760]	Speed: 68.415789 samples/sec	 accuracy=64.682500	 loss=1.449996	 lr=0.001000
Epoch[048] Batch [1299]/[3760]	Speed: 70.316812 samples/sec	 accuracy=64.679087	 loss=1.449573	 lr=0.001000
Epoch[048] Batch [1349]/[3760]	Speed: 70.403185 samples/sec	 accuracy=64.719907	 loss=1.448819	 lr=0.001000
Epoch[048] Batch [1399]/[3760]	Speed: 69.598934 samples/sec	 accuracy=64.669643	 loss=1.448471	 lr=0.001000
Epoch[048] Batch [1449]/[3760]	Speed: 70.464564 samples/sec	 accuracy=64.670259	 loss=1.447912	 lr=0.001000
Epoch[048] Batch [1499]/[3760]	Speed: 69.590589 samples/sec	 accuracy=64.683333	 loss=1.447250	 lr=0.001000
Epoch[048] Batch [1549]/[3760]	Speed: 68.527883 samples/sec	 accuracy=64.675403	 loss=1.448057	 lr=0.001000
Epoch[048] Batch [1599]/[3760]	Speed: 71.178785 samples/sec	 accuracy=64.720703	 loss=1.447176	 lr=0.001000
Epoch[048] Batch [1649]/[3760]	Speed: 66.291614 samples/sec	 accuracy=64.723485	 loss=1.446872	 lr=0.001000
Epoch[048] Batch [1699]/[3760]	Speed: 70.914389 samples/sec	 accuracy=64.718750	 loss=1.445818	 lr=0.001000
Epoch[048] Batch [1749]/[3760]	Speed: 69.600094 samples/sec	 accuracy=64.766071	 loss=1.444848	 lr=0.001000
Epoch[048] Batch [1799]/[3760]	Speed: 70.440398 samples/sec	 accuracy=64.769097	 loss=1.446138	 lr=0.001000
Epoch[048] Batch [1849]/[3760]	Speed: 69.728116 samples/sec	 accuracy=64.764358	 loss=1.446307	 lr=0.001000
Epoch[048] Batch [1899]/[3760]	Speed: 70.308960 samples/sec	 accuracy=64.733553	 loss=1.446370	 lr=0.001000
Epoch[048] Batch [1949]/[3760]	Speed: 69.903060 samples/sec	 accuracy=64.754006	 loss=1.445468	 lr=0.001000
Epoch[048] Batch [1999]/[3760]	Speed: 69.637884 samples/sec	 accuracy=64.750781	 loss=1.445800	 lr=0.001000
Epoch[048] Batch [2049]/[3760]	Speed: 68.668945 samples/sec	 accuracy=64.725610	 loss=1.446375	 lr=0.001000
Epoch[048] Batch [2099]/[3760]	Speed: 67.850776 samples/sec	 accuracy=64.713542	 loss=1.447402	 lr=0.001000
Epoch[048] Batch [2149]/[3760]	Speed: 71.403462 samples/sec	 accuracy=64.732558	 loss=1.446918	 lr=0.001000
Epoch[048] Batch [2199]/[3760]	Speed: 69.559811 samples/sec	 accuracy=64.689631	 loss=1.448030	 lr=0.001000
Epoch[048] Batch [2249]/[3760]	Speed: 70.956874 samples/sec	 accuracy=64.690278	 loss=1.447717	 lr=0.001000
Epoch[048] Batch [2299]/[3760]	Speed: 70.248486 samples/sec	 accuracy=64.695652	 loss=1.447205	 lr=0.001000
Epoch[048] Batch [2349]/[3760]	Speed: 69.286275 samples/sec	 accuracy=64.684840	 loss=1.447424	 lr=0.001000
Epoch[048] Batch [2399]/[3760]	Speed: 69.915098 samples/sec	 accuracy=64.728516	 loss=1.445987	 lr=0.001000
Epoch[048] Batch [2449]/[3760]	Speed: 68.261283 samples/sec	 accuracy=64.747449	 loss=1.445732	 lr=0.001000
Epoch[048] Batch [2499]/[3760]	Speed: 71.391390 samples/sec	 accuracy=64.766875	 loss=1.445744	 lr=0.001000
Epoch[048] Batch [2549]/[3760]	Speed: 69.869103 samples/sec	 accuracy=64.776348	 loss=1.445240	 lr=0.001000
Epoch[048] Batch [2599]/[3760]	Speed: 69.088884 samples/sec	 accuracy=64.758413	 loss=1.446045	 lr=0.001000
Epoch[048] Batch [2649]/[3760]	Speed: 71.396757 samples/sec	 accuracy=64.745873	 loss=1.446456	 lr=0.001000
Epoch[048] Batch [2699]/[3760]	Speed: 69.570693 samples/sec	 accuracy=64.744213	 loss=1.446591	 lr=0.001000
Epoch[048] Batch [2749]/[3760]	Speed: 67.303385 samples/sec	 accuracy=64.733523	 loss=1.446703	 lr=0.001000
Epoch[048] Batch [2799]/[3760]	Speed: 71.272119 samples/sec	 accuracy=64.718192	 loss=1.447108	 lr=0.001000
Epoch[048] Batch [2849]/[3760]	Speed: 70.833108 samples/sec	 accuracy=64.702851	 loss=1.447166	 lr=0.001000
Epoch[048] Batch [2899]/[3760]	Speed: 67.389267 samples/sec	 accuracy=64.710668	 loss=1.447098	 lr=0.001000
Epoch[048] Batch [2949]/[3760]	Speed: 71.496692 samples/sec	 accuracy=64.697564	 loss=1.447548	 lr=0.001000
Epoch[048] Batch [2999]/[3760]	Speed: 70.135226 samples/sec	 accuracy=64.694271	 loss=1.447280	 lr=0.001000
Epoch[048] Batch [3049]/[3760]	Speed: 69.065540 samples/sec	 accuracy=64.675717	 loss=1.447900	 lr=0.001000
Epoch[048] Batch [3099]/[3760]	Speed: 69.951206 samples/sec	 accuracy=64.706653	 loss=1.447044	 lr=0.001000
Epoch[048] Batch [3149]/[3760]	Speed: 69.897397 samples/sec	 accuracy=64.689980	 loss=1.447530	 lr=0.001000
Epoch[048] Batch [3199]/[3760]	Speed: 70.006771 samples/sec	 accuracy=64.698242	 loss=1.447400	 lr=0.001000
Epoch[048] Batch [3249]/[3760]	Speed: 69.799413 samples/sec	 accuracy=64.687981	 loss=1.447683	 lr=0.001000
Epoch[048] Batch [3299]/[3760]	Speed: 70.056778 samples/sec	 accuracy=64.674716	 loss=1.447882	 lr=0.001000
Epoch[048] Batch [3349]/[3760]	Speed: 71.060073 samples/sec	 accuracy=64.676306	 loss=1.447595	 lr=0.001000
Epoch[048] Batch [3399]/[3760]	Speed: 68.967779 samples/sec	 accuracy=64.671875	 loss=1.446972	 lr=0.001000
Epoch[048] Batch [3449]/[3760]	Speed: 69.176532 samples/sec	 accuracy=64.665761	 loss=1.447573	 lr=0.001000
Epoch[048] Batch [3499]/[3760]	Speed: 69.682683 samples/sec	 accuracy=64.688839	 loss=1.446982	 lr=0.001000
Epoch[048] Batch [3549]/[3760]	Speed: 70.273989 samples/sec	 accuracy=64.647447	 loss=1.448104	 lr=0.001000
Epoch[048] Batch [3599]/[3760]	Speed: 70.049982 samples/sec	 accuracy=64.668837	 loss=1.447769	 lr=0.001000
Epoch[048] Batch [3649]/[3760]	Speed: 69.669166 samples/sec	 accuracy=64.657962	 loss=1.448084	 lr=0.001000
Epoch[048] Batch [3699]/[3760]	Speed: 68.743330 samples/sec	 accuracy=64.662162	 loss=1.447896	 lr=0.001000
Epoch[048] Batch [3749]/[3760]	Speed: 77.869149 samples/sec	 accuracy=64.652500	 loss=1.448074	 lr=0.001000
Batch [0049]/[0303]: acc-top1=61.062500 acc-top5=83.343750
Batch [0099]/[0303]: acc-top1=61.156250 acc-top5=83.296875
Batch [0149]/[0303]: acc-top1=61.947917 acc-top5=83.572917
Batch [0199]/[0303]: acc-top1=62.492187 acc-top5=83.843750
Batch [0249]/[0303]: acc-top1=62.631250 acc-top5=84.006250
Batch [0299]/[0303]: acc-top1=62.447917 acc-top5=83.859375
[Epoch 048] training: accuracy=64.655502	 loss=1.447950
[Epoch 048] speed: 69 samples/sec	time cost: 3755.367825
[Epoch 048] validation: acc-top1=62.443276 acc-top5=83.869637 loss=1.627257
Epoch[049] Batch [0049]/[3759]	Speed: 46.312777 samples/sec	 accuracy=64.531250	 loss=1.461223	 lr=0.001000
Epoch[049] Batch [0099]/[3759]	Speed: 69.635380 samples/sec	 accuracy=64.828125	 loss=1.448994	 lr=0.001000
Epoch[049] Batch [0149]/[3759]	Speed: 69.302297 samples/sec	 accuracy=65.302083	 loss=1.438441	 lr=0.001000
Epoch[049] Batch [0199]/[3759]	Speed: 70.668841 samples/sec	 accuracy=65.375000	 loss=1.431898	 lr=0.001000
Epoch[049] Batch [0249]/[3759]	Speed: 70.550404 samples/sec	 accuracy=65.381250	 loss=1.433422	 lr=0.001000
Epoch[049] Batch [0299]/[3759]	Speed: 68.654997 samples/sec	 accuracy=65.395833	 loss=1.421262	 lr=0.001000
Epoch[049] Batch [0349]/[3759]	Speed: 70.185530 samples/sec	 accuracy=65.473214	 loss=1.417758	 lr=0.001000
Epoch[049] Batch [0399]/[3759]	Speed: 69.966967 samples/sec	 accuracy=65.578125	 loss=1.416682	 lr=0.001000
Epoch[049] Batch [0449]/[3759]	Speed: 69.740823 samples/sec	 accuracy=65.559028	 loss=1.419477	 lr=0.001000
Epoch[049] Batch [0499]/[3759]	Speed: 69.795880 samples/sec	 accuracy=65.528125	 loss=1.418786	 lr=0.001000
Epoch[049] Batch [0549]/[3759]	Speed: 68.935464 samples/sec	 accuracy=65.551136	 loss=1.421607	 lr=0.001000
Epoch[049] Batch [0599]/[3759]	Speed: 70.256371 samples/sec	 accuracy=65.632812	 loss=1.417943	 lr=0.001000
Epoch[049] Batch [0649]/[3759]	Speed: 70.239283 samples/sec	 accuracy=65.586538	 loss=1.419135	 lr=0.001000
Epoch[049] Batch [0699]/[3759]	Speed: 68.367628 samples/sec	 accuracy=65.531250	 loss=1.421691	 lr=0.001000
Epoch[049] Batch [0749]/[3759]	Speed: 70.141742 samples/sec	 accuracy=65.460417	 loss=1.421653	 lr=0.001000
Epoch[049] Batch [0799]/[3759]	Speed: 70.708382 samples/sec	 accuracy=65.355469	 loss=1.426961	 lr=0.001000
Epoch[049] Batch [0849]/[3759]	Speed: 69.707960 samples/sec	 accuracy=65.349265	 loss=1.425745	 lr=0.001000
Epoch[049] Batch [0899]/[3759]	Speed: 69.282089 samples/sec	 accuracy=65.272569	 loss=1.428045	 lr=0.001000
Epoch[049] Batch [0949]/[3759]	Speed: 69.807346 samples/sec	 accuracy=65.261513	 loss=1.429737	 lr=0.001000
Epoch[049] Batch [0999]/[3759]	Speed: 70.030832 samples/sec	 accuracy=65.201562	 loss=1.430772	 lr=0.001000
Epoch[049] Batch [1049]/[3759]	Speed: 68.221375 samples/sec	 accuracy=65.194940	 loss=1.429150	 lr=0.001000
Epoch[049] Batch [1099]/[3759]	Speed: 70.990547 samples/sec	 accuracy=65.196023	 loss=1.429820	 lr=0.001000
Epoch[049] Batch [1149]/[3759]	Speed: 69.668035 samples/sec	 accuracy=65.148098	 loss=1.430127	 lr=0.001000
Epoch[049] Batch [1199]/[3759]	Speed: 70.875019 samples/sec	 accuracy=65.089844	 loss=1.431740	 lr=0.001000
Epoch[049] Batch [1249]/[3759]	Speed: 69.499145 samples/sec	 accuracy=65.143750	 loss=1.430501	 lr=0.001000
Epoch[049] Batch [1299]/[3759]	Speed: 70.142243 samples/sec	 accuracy=65.162260	 loss=1.430244	 lr=0.001000
Epoch[049] Batch [1349]/[3759]	Speed: 69.527450 samples/sec	 accuracy=65.164352	 loss=1.429497	 lr=0.001000
Epoch[049] Batch [1399]/[3759]	Speed: 68.431915 samples/sec	 accuracy=65.157366	 loss=1.431078	 lr=0.001000
Epoch[049] Batch [1449]/[3759]	Speed: 70.231395 samples/sec	 accuracy=65.150862	 loss=1.431751	 lr=0.001000
Epoch[049] Batch [1499]/[3759]	Speed: 70.191958 samples/sec	 accuracy=65.167708	 loss=1.431137	 lr=0.001000
Epoch[049] Batch [1549]/[3759]	Speed: 69.868463 samples/sec	 accuracy=65.192540	 loss=1.429916	 lr=0.001000
Epoch[049] Batch [1599]/[3759]	Speed: 70.251658 samples/sec	 accuracy=65.192383	 loss=1.429937	 lr=0.001000
Epoch[049] Batch [1649]/[3759]	Speed: 69.978307 samples/sec	 accuracy=65.164773	 loss=1.430616	 lr=0.001000
Epoch[049] Batch [1699]/[3759]	Speed: 68.524604 samples/sec	 accuracy=65.159007	 loss=1.430289	 lr=0.001000
Epoch[049] Batch [1749]/[3759]	Speed: 69.044195 samples/sec	 accuracy=65.118750	 loss=1.431592	 lr=0.001000
Epoch[049] Batch [1799]/[3759]	Speed: 71.126886 samples/sec	 accuracy=65.123264	 loss=1.430516	 lr=0.001000
Epoch[049] Batch [1849]/[3759]	Speed: 70.084251 samples/sec	 accuracy=65.128378	 loss=1.429996	 lr=0.001000
Epoch[049] Batch [1899]/[3759]	Speed: 70.280599 samples/sec	 accuracy=65.135691	 loss=1.429565	 lr=0.001000
Epoch[049] Batch [1949]/[3759]	Speed: 68.385417 samples/sec	 accuracy=65.125000	 loss=1.430997	 lr=0.001000
Epoch[049] Batch [1999]/[3759]	Speed: 69.239117 samples/sec	 accuracy=65.092969	 loss=1.432508	 lr=0.001000
Epoch[049] Batch [2049]/[3759]	Speed: 70.360381 samples/sec	 accuracy=65.091463	 loss=1.432768	 lr=0.001000
Epoch[049] Batch [2099]/[3759]	Speed: 69.015531 samples/sec	 accuracy=65.056548	 loss=1.434096	 lr=0.001000
Epoch[049] Batch [2149]/[3759]	Speed: 70.834591 samples/sec	 accuracy=65.029070	 loss=1.434701	 lr=0.001000
Epoch[049] Batch [2199]/[3759]	Speed: 70.086535 samples/sec	 accuracy=65.054688	 loss=1.433812	 lr=0.001000
Epoch[049] Batch [2249]/[3759]	Speed: 70.233059 samples/sec	 accuracy=65.019444	 loss=1.435177	 lr=0.001000
Epoch[049] Batch [2299]/[3759]	Speed: 68.825928 samples/sec	 accuracy=64.985734	 loss=1.436366	 lr=0.001000
Epoch[049] Batch [2349]/[3759]	Speed: 68.296507 samples/sec	 accuracy=64.962101	 loss=1.436456	 lr=0.001000
Epoch[049] Batch [2399]/[3759]	Speed: 71.632253 samples/sec	 accuracy=65.005208	 loss=1.435603	 lr=0.001000
Epoch[049] Batch [2449]/[3759]	Speed: 69.134607 samples/sec	 accuracy=65.018495	 loss=1.434932	 lr=0.001000
Epoch[049] Batch [2499]/[3759]	Speed: 68.327073 samples/sec	 accuracy=65.028125	 loss=1.435784	 lr=0.001000
Epoch[049] Batch [2549]/[3759]	Speed: 70.912394 samples/sec	 accuracy=65.022672	 loss=1.436261	 lr=0.001000
Epoch[049] Batch [2599]/[3759]	Speed: 70.037297 samples/sec	 accuracy=65.003005	 loss=1.436871	 lr=0.001000
Epoch[049] Batch [2649]/[3759]	Speed: 70.636478 samples/sec	 accuracy=65.001769	 loss=1.436617	 lr=0.001000
Epoch[049] Batch [2699]/[3759]	Speed: 68.178842 samples/sec	 accuracy=65.000579	 loss=1.435912	 lr=0.001000
Epoch[049] Batch [2749]/[3759]	Speed: 69.582018 samples/sec	 accuracy=64.986364	 loss=1.436712	 lr=0.001000
Epoch[049] Batch [2799]/[3759]	Speed: 70.137417 samples/sec	 accuracy=64.977121	 loss=1.437056	 lr=0.001000
Epoch[049] Batch [2849]/[3759]	Speed: 71.005031 samples/sec	 accuracy=64.995614	 loss=1.435855	 lr=0.001000
Epoch[049] Batch [2899]/[3759]	Speed: 68.871010 samples/sec	 accuracy=64.989763	 loss=1.435710	 lr=0.001000
Epoch[049] Batch [2949]/[3759]	Speed: 70.532570 samples/sec	 accuracy=64.995233	 loss=1.435597	 lr=0.001000
Epoch[049] Batch [2999]/[3759]	Speed: 68.191518 samples/sec	 accuracy=64.998437	 loss=1.435296	 lr=0.001000
Epoch[049] Batch [3049]/[3759]	Speed: 70.469774 samples/sec	 accuracy=64.999488	 loss=1.435507	 lr=0.001000
Epoch[049] Batch [3099]/[3759]	Speed: 69.707344 samples/sec	 accuracy=64.993448	 loss=1.435261	 lr=0.001000
Epoch[049] Batch [3149]/[3759]	Speed: 69.946481 samples/sec	 accuracy=65.002480	 loss=1.435104	 lr=0.001000
Epoch[049] Batch [3199]/[3759]	Speed: 69.182157 samples/sec	 accuracy=64.996094	 loss=1.435737	 lr=0.001000
Epoch[049] Batch [3249]/[3759]	Speed: 71.170491 samples/sec	 accuracy=64.991346	 loss=1.435637	 lr=0.001000
Epoch[049] Batch [3299]/[3759]	Speed: 69.507432 samples/sec	 accuracy=64.989110	 loss=1.435548	 lr=0.001000
Epoch[049] Batch [3349]/[3759]	Speed: 69.083171 samples/sec	 accuracy=64.987407	 loss=1.435682	 lr=0.001000
Epoch[049] Batch [3399]/[3759]	Speed: 69.375088 samples/sec	 accuracy=64.976103	 loss=1.435908	 lr=0.001000
Epoch[049] Batch [3449]/[3759]	Speed: 69.862007 samples/sec	 accuracy=64.956069	 loss=1.436770	 lr=0.001000
Epoch[049] Batch [3499]/[3759]	Speed: 70.334855 samples/sec	 accuracy=64.950000	 loss=1.437315	 lr=0.001000
Epoch[049] Batch [3549]/[3759]	Speed: 70.805389 samples/sec	 accuracy=64.937060	 loss=1.437723	 lr=0.001000
Epoch[049] Batch [3599]/[3759]	Speed: 69.726249 samples/sec	 accuracy=64.939236	 loss=1.438286	 lr=0.001000
Epoch[049] Batch [3649]/[3759]	Speed: 69.475490 samples/sec	 accuracy=64.947774	 loss=1.438064	 lr=0.001000
Epoch[049] Batch [3699]/[3759]	Speed: 70.339876 samples/sec	 accuracy=64.967483	 loss=1.437368	 lr=0.001000
Epoch[049] Batch [3749]/[3759]	Speed: 77.958134 samples/sec	 accuracy=64.970000	 loss=1.437690	 lr=0.001000
Batch [0049]/[0303]: acc-top1=61.281250 acc-top5=83.312500
Batch [0099]/[0303]: acc-top1=61.250000 acc-top5=83.203125
Batch [0149]/[0303]: acc-top1=62.239583 acc-top5=83.604167
Batch [0199]/[0303]: acc-top1=62.554687 acc-top5=83.984375
Batch [0249]/[0303]: acc-top1=62.593750 acc-top5=84.118750
Batch [0299]/[0303]: acc-top1=62.354167 acc-top5=83.932292
[Epoch 049] training: accuracy=64.967827	 loss=1.438005
[Epoch 049] speed: 69 samples/sec	time cost: 3753.296234
[Epoch 049] validation: acc-top1=62.314356 acc-top5=83.946988 loss=1.631884
Epoch[050] Batch [0049]/[3760]	Speed: 48.737868 samples/sec	 accuracy=65.500000	 loss=1.411852	 lr=0.001000
Epoch[050] Batch [0099]/[3760]	Speed: 68.251219 samples/sec	 accuracy=64.843750	 loss=1.429974	 lr=0.001000
Epoch[050] Batch [0149]/[3760]	Speed: 67.162997 samples/sec	 accuracy=65.083333	 loss=1.422392	 lr=0.001000
Epoch[050] Batch [0199]/[3760]	Speed: 70.943552 samples/sec	 accuracy=65.195312	 loss=1.421866	 lr=0.001000
Epoch[050] Batch [0249]/[3760]	Speed: 70.790892 samples/sec	 accuracy=65.725000	 loss=1.404986	 lr=0.001000
Epoch[050] Batch [0299]/[3760]	Speed: 67.549495 samples/sec	 accuracy=65.942708	 loss=1.400932	 lr=0.001000
Epoch[050] Batch [0349]/[3760]	Speed: 71.916953 samples/sec	 accuracy=65.986607	 loss=1.401312	 lr=0.001000
Epoch[050] Batch [0399]/[3760]	Speed: 69.313136 samples/sec	 accuracy=65.789062	 loss=1.404245	 lr=0.001000
Epoch[050] Batch [0449]/[3760]	Speed: 70.083759 samples/sec	 accuracy=65.788194	 loss=1.406702	 lr=0.001000
Epoch[050] Batch [0499]/[3760]	Speed: 68.099547 samples/sec	 accuracy=65.668750	 loss=1.409397	 lr=0.001000
Epoch[050] Batch [0549]/[3760]	Speed: 69.948006 samples/sec	 accuracy=65.642045	 loss=1.412750	 lr=0.001000
Epoch[050] Batch [0599]/[3760]	Speed: 70.435079 samples/sec	 accuracy=65.544271	 loss=1.412186	 lr=0.001000
Epoch[050] Batch [0649]/[3760]	Speed: 70.287477 samples/sec	 accuracy=65.605769	 loss=1.409681	 lr=0.001000
Epoch[050] Batch [0699]/[3760]	Speed: 70.807540 samples/sec	 accuracy=65.544643	 loss=1.412102	 lr=0.001000
Epoch[050] Batch [0749]/[3760]	Speed: 68.343493 samples/sec	 accuracy=65.481250	 loss=1.413302	 lr=0.001000
Epoch[050] Batch [0799]/[3760]	Speed: 70.994266 samples/sec	 accuracy=65.392578	 loss=1.416344	 lr=0.001000
Epoch[050] Batch [0849]/[3760]	Speed: 68.909327 samples/sec	 accuracy=65.319853	 loss=1.417767	 lr=0.001000
Epoch[050] Batch [0899]/[3760]	Speed: 70.888173 samples/sec	 accuracy=65.310764	 loss=1.416914	 lr=0.001000
Epoch[050] Batch [0949]/[3760]	Speed: 69.838327 samples/sec	 accuracy=65.220395	 loss=1.420708	 lr=0.001000
Epoch[050] Batch [0999]/[3760]	Speed: 68.332742 samples/sec	 accuracy=65.246875	 loss=1.418239	 lr=0.001000
Epoch[050] Batch [1049]/[3760]	Speed: 71.018543 samples/sec	 accuracy=65.238095	 loss=1.419613	 lr=0.001000
Epoch[050] Batch [1099]/[3760]	Speed: 69.004048 samples/sec	 accuracy=65.245739	 loss=1.419886	 lr=0.001000
Epoch[050] Batch [1149]/[3760]	Speed: 70.113797 samples/sec	 accuracy=65.305707	 loss=1.418535	 lr=0.001000
Epoch[050] Batch [1199]/[3760]	Speed: 69.579359 samples/sec	 accuracy=65.278646	 loss=1.419785	 lr=0.001000
Epoch[050] Batch [1249]/[3760]	Speed: 69.921571 samples/sec	 accuracy=65.276250	 loss=1.420242	 lr=0.001000
Epoch[050] Batch [1299]/[3760]	Speed: 64.867492 samples/sec	 accuracy=65.271635	 loss=1.421089	 lr=0.001000
Epoch[050] Batch [1349]/[3760]	Speed: 71.603039 samples/sec	 accuracy=65.281250	 loss=1.420376	 lr=0.001000
Epoch[050] Batch [1399]/[3760]	Speed: 70.914248 samples/sec	 accuracy=65.281250	 loss=1.420201	 lr=0.001000
Epoch[050] Batch [1449]/[3760]	Speed: 67.073608 samples/sec	 accuracy=65.332974	 loss=1.418156	 lr=0.001000
Epoch[050] Batch [1499]/[3760]	Speed: 72.138974 samples/sec	 accuracy=65.286458	 loss=1.419405	 lr=0.001000
Epoch[050] Batch [1549]/[3760]	Speed: 69.515121 samples/sec	 accuracy=65.258065	 loss=1.421031	 lr=0.001000
Epoch[050] Batch [1599]/[3760]	Speed: 69.094398 samples/sec	 accuracy=65.270508	 loss=1.421233	 lr=0.001000
Epoch[050] Batch [1649]/[3760]	Speed: 68.989812 samples/sec	 accuracy=65.285038	 loss=1.421110	 lr=0.001000
Epoch[050] Batch [1699]/[3760]	Speed: 70.249707 samples/sec	 accuracy=65.279412	 loss=1.421728	 lr=0.001000
Epoch[050] Batch [1749]/[3760]	Speed: 69.582944 samples/sec	 accuracy=65.277679	 loss=1.421186	 lr=0.001000
Epoch[050] Batch [1799]/[3760]	Speed: 69.318130 samples/sec	 accuracy=65.280382	 loss=1.420007	 lr=0.001000
Epoch[050] Batch [1849]/[3760]	Speed: 68.927654 samples/sec	 accuracy=65.288007	 loss=1.419853	 lr=0.001000
Epoch[050] Batch [1899]/[3760]	Speed: 68.265028 samples/sec	 accuracy=65.273026	 loss=1.420445	 lr=0.001000
Epoch[050] Batch [1949]/[3760]	Speed: 68.841935 samples/sec	 accuracy=65.256410	 loss=1.421246	 lr=0.001000
Epoch[050] Batch [1999]/[3760]	Speed: 72.898198 samples/sec	 accuracy=65.234375	 loss=1.421294	 lr=0.001000
Epoch[050] Batch [2049]/[3760]	Speed: 67.146834 samples/sec	 accuracy=65.240854	 loss=1.420527	 lr=0.001000
Epoch[050] Batch [2099]/[3760]	Speed: 68.172852 samples/sec	 accuracy=65.221726	 loss=1.421516	 lr=0.001000
Epoch[050] Batch [2149]/[3760]	Speed: 74.265909 samples/sec	 accuracy=65.199855	 loss=1.422663	 lr=0.001000
Epoch[050] Batch [2199]/[3760]	Speed: 69.243027 samples/sec	 accuracy=65.165483	 loss=1.422926	 lr=0.001000
Epoch[050] Batch [2249]/[3760]	Speed: 67.815632 samples/sec	 accuracy=65.140278	 loss=1.423537	 lr=0.001000
Epoch[050] Batch [2299]/[3760]	Speed: 71.488182 samples/sec	 accuracy=65.148777	 loss=1.423513	 lr=0.001000
Epoch[050] Batch [2349]/[3760]	Speed: 69.444163 samples/sec	 accuracy=65.111702	 loss=1.424402	 lr=0.001000
Epoch[050] Batch [2399]/[3760]	Speed: 69.918213 samples/sec	 accuracy=65.125651	 loss=1.424468	 lr=0.001000
Epoch[050] Batch [2449]/[3760]	Speed: 68.621211 samples/sec	 accuracy=65.147959	 loss=1.424398	 lr=0.001000
Epoch[050] Batch [2499]/[3760]	Speed: 69.468634 samples/sec	 accuracy=65.131875	 loss=1.424425	 lr=0.001000
Epoch[050] Batch [2549]/[3760]	Speed: 68.365501 samples/sec	 accuracy=65.143382	 loss=1.423313	 lr=0.001000
Epoch[050] Batch [2599]/[3760]	Speed: 70.532454 samples/sec	 accuracy=65.146635	 loss=1.423564	 lr=0.001000
Epoch[050] Batch [2649]/[3760]	Speed: 69.333035 samples/sec	 accuracy=65.145637	 loss=1.422812	 lr=0.001000
Epoch[050] Batch [2699]/[3760]	Speed: 69.266751 samples/sec	 accuracy=65.159722	 loss=1.422856	 lr=0.001000
Epoch[050] Batch [2749]/[3760]	Speed: 69.479395 samples/sec	 accuracy=65.163636	 loss=1.423177	 lr=0.001000
Epoch[050] Batch [2799]/[3760]	Speed: 70.147720 samples/sec	 accuracy=65.192522	 loss=1.422555	 lr=0.001000
Epoch[050] Batch [2849]/[3760]	Speed: 66.844784 samples/sec	 accuracy=65.199561	 loss=1.422412	 lr=0.001000
Epoch[050] Batch [2899]/[3760]	Speed: 70.872371 samples/sec	 accuracy=65.169720	 loss=1.423484	 lr=0.001000
Epoch[050] Batch [2949]/[3760]	Speed: 70.476029 samples/sec	 accuracy=65.172140	 loss=1.423650	 lr=0.001000
Epoch[050] Batch [2999]/[3760]	Speed: 68.346564 samples/sec	 accuracy=65.170313	 loss=1.423251	 lr=0.001000
Epoch[050] Batch [3049]/[3760]	Speed: 67.523546 samples/sec	 accuracy=65.171107	 loss=1.423861	 lr=0.001000
Epoch[050] Batch [3099]/[3760]	Speed: 71.525301 samples/sec	 accuracy=65.171371	 loss=1.424270	 lr=0.001000
Epoch[050] Batch [3149]/[3760]	Speed: 69.553866 samples/sec	 accuracy=65.151786	 loss=1.424720	 lr=0.001000
Epoch[050] Batch [3199]/[3760]	Speed: 69.935324 samples/sec	 accuracy=65.166016	 loss=1.424585	 lr=0.001000
Epoch[050] Batch [3249]/[3760]	Speed: 68.413779 samples/sec	 accuracy=65.158654	 loss=1.424955	 lr=0.001000
Epoch[050] Batch [3299]/[3760]	Speed: 70.436657 samples/sec	 accuracy=65.144413	 loss=1.425166	 lr=0.001000
Epoch[050] Batch [3349]/[3760]	Speed: 67.781767 samples/sec	 accuracy=65.136660	 loss=1.426132	 lr=0.001000
Epoch[050] Batch [3399]/[3760]	Speed: 70.186242 samples/sec	 accuracy=65.126838	 loss=1.426254	 lr=0.001000
Epoch[050] Batch [3449]/[3760]	Speed: 67.289427 samples/sec	 accuracy=65.129982	 loss=1.426297	 lr=0.001000
Epoch[050] Batch [3499]/[3760]	Speed: 72.366969 samples/sec	 accuracy=65.135268	 loss=1.425770	 lr=0.001000
Epoch[050] Batch [3549]/[3760]	Speed: 70.030918 samples/sec	 accuracy=65.124120	 loss=1.426080	 lr=0.001000
Epoch[050] Batch [3599]/[3760]	Speed: 69.091230 samples/sec	 accuracy=65.114583	 loss=1.426108	 lr=0.001000
Epoch[050] Batch [3649]/[3760]	Speed: 70.318322 samples/sec	 accuracy=65.123288	 loss=1.425591	 lr=0.001000
Epoch[050] Batch [3699]/[3760]	Speed: 67.865356 samples/sec	 accuracy=65.122889	 loss=1.425918	 lr=0.001000
Epoch[050] Batch [3749]/[3760]	Speed: 77.809180 samples/sec	 accuracy=65.120833	 loss=1.426469	 lr=0.001000
Batch [0049]/[0303]: acc-top1=61.500000 acc-top5=83.187500
Batch [0099]/[0303]: acc-top1=61.281250 acc-top5=83.250000
Batch [0149]/[0303]: acc-top1=62.333333 acc-top5=83.760417
Batch [0199]/[0303]: acc-top1=62.562500 acc-top5=84.101562
Batch [0249]/[0303]: acc-top1=62.700000 acc-top5=84.281250
Batch [0299]/[0303]: acc-top1=62.526042 acc-top5=84.135417
[Epoch 050] training: accuracy=65.113863	 loss=1.426774
[Epoch 050] speed: 69 samples/sec	time cost: 3760.867624
[Epoch 050] validation: acc-top1=62.500000 acc-top5=84.142946 loss=1.624335
Epoch[051] Batch [0049]/[3760]	Speed: 48.045240 samples/sec	 accuracy=65.625000	 loss=1.385030	 lr=0.001000
Epoch[051] Batch [0099]/[3760]	Speed: 67.734782 samples/sec	 accuracy=66.250000	 loss=1.377427	 lr=0.001000
Epoch[051] Batch [0149]/[3760]	Speed: 69.613115 samples/sec	 accuracy=65.937500	 loss=1.394178	 lr=0.001000
Epoch[051] Batch [0199]/[3760]	Speed: 69.034859 samples/sec	 accuracy=65.867188	 loss=1.401529	 lr=0.001000
Epoch[051] Batch [0249]/[3760]	Speed: 68.326646 samples/sec	 accuracy=66.093750	 loss=1.396171	 lr=0.001000
Epoch[051] Batch [0299]/[3760]	Speed: 70.066366 samples/sec	 accuracy=65.838542	 loss=1.403288	 lr=0.001000
Epoch[051] Batch [0349]/[3760]	Speed: 70.889775 samples/sec	 accuracy=65.848214	 loss=1.402036	 lr=0.001000
Epoch[051] Batch [0399]/[3760]	Speed: 69.241682 samples/sec	 accuracy=65.894531	 loss=1.397397	 lr=0.001000
Epoch[051] Batch [0449]/[3760]	Speed: 69.426291 samples/sec	 accuracy=65.767361	 loss=1.402644	 lr=0.001000
Epoch[051] Batch [0499]/[3760]	Speed: 70.173568 samples/sec	 accuracy=65.650000	 loss=1.406768	 lr=0.001000
Epoch[051] Batch [0549]/[3760]	Speed: 68.189699 samples/sec	 accuracy=65.661932	 loss=1.402630	 lr=0.001000
Epoch[051] Batch [0599]/[3760]	Speed: 69.818781 samples/sec	 accuracy=65.552083	 loss=1.406007	 lr=0.001000
Epoch[051] Batch [0649]/[3760]	Speed: 67.308958 samples/sec	 accuracy=65.576923	 loss=1.406922	 lr=0.001000
Epoch[051] Batch [0699]/[3760]	Speed: 71.942815 samples/sec	 accuracy=65.584821	 loss=1.409122	 lr=0.001000
Epoch[051] Batch [0749]/[3760]	Speed: 70.262682 samples/sec	 accuracy=65.600000	 loss=1.408686	 lr=0.001000
Epoch[051] Batch [0799]/[3760]	Speed: 68.178132 samples/sec	 accuracy=65.589844	 loss=1.410554	 lr=0.001000
Epoch[051] Batch [0849]/[3760]	Speed: 71.467890 samples/sec	 accuracy=65.516544	 loss=1.412217	 lr=0.001000
Epoch[051] Batch [0899]/[3760]	Speed: 68.561163 samples/sec	 accuracy=65.569444	 loss=1.409821	 lr=0.001000
Epoch[051] Batch [0949]/[3760]	Speed: 70.559629 samples/sec	 accuracy=65.634868	 loss=1.406329	 lr=0.001000
Epoch[051] Batch [0999]/[3760]	Speed: 67.232529 samples/sec	 accuracy=65.667187	 loss=1.405154	 lr=0.001000
Epoch[051] Batch [1049]/[3760]	Speed: 71.256056 samples/sec	 accuracy=65.629464	 loss=1.404810	 lr=0.001000
Epoch[051] Batch [1099]/[3760]	Speed: 70.814293 samples/sec	 accuracy=65.566761	 loss=1.408144	 lr=0.001000
Epoch[051] Batch [1149]/[3760]	Speed: 68.592932 samples/sec	 accuracy=65.538043	 loss=1.407946	 lr=0.001000
Epoch[051] Batch [1199]/[3760]	Speed: 70.382190 samples/sec	 accuracy=65.507812	 loss=1.409756	 lr=0.001000
Epoch[051] Batch [1249]/[3760]	Speed: 69.644785 samples/sec	 accuracy=65.543750	 loss=1.408208	 lr=0.001000
Epoch[051] Batch [1299]/[3760]	Speed: 69.181809 samples/sec	 accuracy=65.575721	 loss=1.407833	 lr=0.001000
Epoch[051] Batch [1349]/[3760]	Speed: 69.460199 samples/sec	 accuracy=65.616898	 loss=1.407503	 lr=0.001000
Epoch[051] Batch [1399]/[3760]	Speed: 69.390339 samples/sec	 accuracy=65.566964	 loss=1.408828	 lr=0.001000
Epoch[051] Batch [1449]/[3760]	Speed: 70.884634 samples/sec	 accuracy=65.544181	 loss=1.408838	 lr=0.001000
Epoch[051] Batch [1499]/[3760]	Speed: 70.095091 samples/sec	 accuracy=65.517708	 loss=1.410137	 lr=0.001000
Epoch[051] Batch [1549]/[3760]	Speed: 69.194446 samples/sec	 accuracy=65.496976	 loss=1.410236	 lr=0.001000
Epoch[051] Batch [1599]/[3760]	Speed: 70.815124 samples/sec	 accuracy=65.541016	 loss=1.409227	 lr=0.001000
Epoch[051] Batch [1649]/[3760]	Speed: 69.782440 samples/sec	 accuracy=65.529356	 loss=1.410086	 lr=0.001000
Epoch[051] Batch [1699]/[3760]	Speed: 67.564659 samples/sec	 accuracy=65.505515	 loss=1.410508	 lr=0.001000
Epoch[051] Batch [1749]/[3760]	Speed: 71.249999 samples/sec	 accuracy=65.474107	 loss=1.411718	 lr=0.001000
Epoch[051] Batch [1799]/[3760]	Speed: 71.034629 samples/sec	 accuracy=65.460938	 loss=1.411597	 lr=0.001000
Epoch[051] Batch [1849]/[3760]	Speed: 69.372616 samples/sec	 accuracy=65.407939	 loss=1.412466	 lr=0.001000
Epoch[051] Batch [1899]/[3760]	Speed: 68.149882 samples/sec	 accuracy=65.386513	 loss=1.413386	 lr=0.001000
Epoch[051] Batch [1949]/[3760]	Speed: 70.282370 samples/sec	 accuracy=65.348558	 loss=1.415603	 lr=0.001000
Epoch[051] Batch [1999]/[3760]	Speed: 70.200975 samples/sec	 accuracy=65.360938	 loss=1.415331	 lr=0.001000
Epoch[051] Batch [2049]/[3760]	Speed: 69.197089 samples/sec	 accuracy=65.378049	 loss=1.415250	 lr=0.001000
Epoch[051] Batch [2099]/[3760]	Speed: 70.300055 samples/sec	 accuracy=65.357143	 loss=1.415501	 lr=0.001000
Epoch[051] Batch [2149]/[3760]	Speed: 67.867531 samples/sec	 accuracy=65.335029	 loss=1.415566	 lr=0.001000
Epoch[051] Batch [2199]/[3760]	Speed: 70.985456 samples/sec	 accuracy=65.362926	 loss=1.414985	 lr=0.001000
Epoch[051] Batch [2249]/[3760]	Speed: 69.777637 samples/sec	 accuracy=65.366667	 loss=1.414675	 lr=0.001000
Epoch[051] Batch [2299]/[3760]	Speed: 70.501030 samples/sec	 accuracy=65.368207	 loss=1.414568	 lr=0.001000
Epoch[051] Batch [2349]/[3760]	Speed: 69.494001 samples/sec	 accuracy=65.375000	 loss=1.414213	 lr=0.001000
Epoch[051] Batch [2399]/[3760]	Speed: 68.794106 samples/sec	 accuracy=65.400391	 loss=1.414093	 lr=0.001000
Epoch[051] Batch [2449]/[3760]	Speed: 70.888252 samples/sec	 accuracy=65.404974	 loss=1.414103	 lr=0.001000
Epoch[051] Batch [2499]/[3760]	Speed: 68.929732 samples/sec	 accuracy=65.423125	 loss=1.414344	 lr=0.001000
Epoch[051] Batch [2549]/[3760]	Speed: 69.992915 samples/sec	 accuracy=65.425245	 loss=1.414372	 lr=0.001000
Epoch[051] Batch [2599]/[3760]	Speed: 68.844047 samples/sec	 accuracy=65.401442	 loss=1.415147	 lr=0.001000
Epoch[051] Batch [2649]/[3760]	Speed: 70.208430 samples/sec	 accuracy=65.405660	 loss=1.414051	 lr=0.001000
Epoch[051] Batch [2699]/[3760]	Speed: 70.889576 samples/sec	 accuracy=65.377315	 loss=1.415100	 lr=0.001000
Epoch[051] Batch [2749]/[3760]	Speed: 69.273033 samples/sec	 accuracy=65.367614	 loss=1.415641	 lr=0.001000
Epoch[051] Batch [2799]/[3760]	Speed: 69.725790 samples/sec	 accuracy=65.352679	 loss=1.416030	 lr=0.001000
Epoch[051] Batch [2849]/[3760]	Speed: 70.735122 samples/sec	 accuracy=65.365132	 loss=1.415800	 lr=0.001000
Epoch[051] Batch [2899]/[3760]	Speed: 69.602742 samples/sec	 accuracy=65.355603	 loss=1.416614	 lr=0.001000
Epoch[051] Batch [2949]/[3760]	Speed: 69.069520 samples/sec	 accuracy=65.381886	 loss=1.416540	 lr=0.001000
Epoch[051] Batch [2999]/[3760]	Speed: 70.786463 samples/sec	 accuracy=65.346354	 loss=1.417952	 lr=0.001000
Epoch[051] Batch [3049]/[3760]	Speed: 68.548360 samples/sec	 accuracy=65.346824	 loss=1.417643	 lr=0.001000
Epoch[051] Batch [3099]/[3760]	Speed: 68.331429 samples/sec	 accuracy=65.349294	 loss=1.417584	 lr=0.001000
Epoch[051] Batch [3149]/[3760]	Speed: 71.300689 samples/sec	 accuracy=65.344742	 loss=1.417827	 lr=0.001000
Epoch[051] Batch [3199]/[3760]	Speed: 70.678581 samples/sec	 accuracy=65.335449	 loss=1.418007	 lr=0.001000
Epoch[051] Batch [3249]/[3760]	Speed: 69.528234 samples/sec	 accuracy=65.313942	 loss=1.418546	 lr=0.001000
Epoch[051] Batch [3299]/[3760]	Speed: 67.779715 samples/sec	 accuracy=65.326705	 loss=1.418392	 lr=0.001000
Epoch[051] Batch [3349]/[3760]	Speed: 70.642333 samples/sec	 accuracy=65.316698	 loss=1.419192	 lr=0.001000
Epoch[051] Batch [3399]/[3760]	Speed: 69.193891 samples/sec	 accuracy=65.325827	 loss=1.419271	 lr=0.001000
Epoch[051] Batch [3449]/[3760]	Speed: 70.600756 samples/sec	 accuracy=65.327446	 loss=1.419170	 lr=0.001000
Epoch[051] Batch [3499]/[3760]	Speed: 70.659787 samples/sec	 accuracy=65.333482	 loss=1.419373	 lr=0.001000
Epoch[051] Batch [3549]/[3760]	Speed: 69.287985 samples/sec	 accuracy=65.329225	 loss=1.419577	 lr=0.001000
Epoch[051] Batch [3599]/[3760]	Speed: 69.703143 samples/sec	 accuracy=65.325087	 loss=1.419758	 lr=0.001000
Epoch[051] Batch [3649]/[3760]	Speed: 69.304398 samples/sec	 accuracy=65.318493	 loss=1.420096	 lr=0.001000
Epoch[051] Batch [3699]/[3760]	Speed: 68.755597 samples/sec	 accuracy=65.326858	 loss=1.419274	 lr=0.001000
Epoch[051] Batch [3749]/[3760]	Speed: 77.471270 samples/sec	 accuracy=65.329167	 loss=1.419476	 lr=0.001000
Batch [0049]/[0303]: acc-top1=61.593750 acc-top5=83.531250
Batch [0099]/[0303]: acc-top1=61.703125 acc-top5=83.250000
Batch [0149]/[0303]: acc-top1=62.500000 acc-top5=83.739583
Batch [0199]/[0303]: acc-top1=62.812500 acc-top5=84.070312
Batch [0249]/[0303]: acc-top1=62.850000 acc-top5=84.275000
Batch [0299]/[0303]: acc-top1=62.598958 acc-top5=84.140625
[Epoch 051] training: accuracy=65.330369	 loss=1.419391
[Epoch 051] speed: 69 samples/sec	time cost: 3754.626061
[Epoch 051] validation: acc-top1=62.572195 acc-top5=84.153259 loss=1.619314
Epoch[052] Batch [0049]/[3759]	Speed: 47.460935 samples/sec	 accuracy=65.500000	 loss=1.386121	 lr=0.001000
Epoch[052] Batch [0099]/[3759]	Speed: 70.110873 samples/sec	 accuracy=65.781250	 loss=1.380632	 lr=0.001000
Epoch[052] Batch [0149]/[3759]	Speed: 68.699757 samples/sec	 accuracy=65.593750	 loss=1.384657	 lr=0.001000
Epoch[052] Batch [0199]/[3759]	Speed: 70.349755 samples/sec	 accuracy=65.843750	 loss=1.388002	 lr=0.001000
Epoch[052] Batch [0249]/[3759]	Speed: 69.976913 samples/sec	 accuracy=65.725000	 loss=1.389013	 lr=0.001000
Epoch[052] Batch [0299]/[3759]	Speed: 70.047947 samples/sec	 accuracy=65.796875	 loss=1.396624	 lr=0.001000
Epoch[052] Batch [0349]/[3759]	Speed: 70.340408 samples/sec	 accuracy=65.705357	 loss=1.399291	 lr=0.001000
Epoch[052] Batch [0399]/[3759]	Speed: 68.725997 samples/sec	 accuracy=65.472656	 loss=1.405887	 lr=0.001000
Epoch[052] Batch [0449]/[3759]	Speed: 69.927809 samples/sec	 accuracy=65.479167	 loss=1.403210	 lr=0.001000
Epoch[052] Batch [0499]/[3759]	Speed: 69.462554 samples/sec	 accuracy=65.506250	 loss=1.401633	 lr=0.001000
Epoch[052] Batch [0549]/[3759]	Speed: 70.007856 samples/sec	 accuracy=65.642045	 loss=1.397306	 lr=0.001000
Epoch[052] Batch [0599]/[3759]	Speed: 70.746154 samples/sec	 accuracy=65.614583	 loss=1.397155	 lr=0.001000
Epoch[052] Batch [0649]/[3759]	Speed: 68.900749 samples/sec	 accuracy=65.617788	 loss=1.396989	 lr=0.001000
Epoch[052] Batch [0699]/[3759]	Speed: 69.854780 samples/sec	 accuracy=65.558036	 loss=1.400093	 lr=0.001000
Epoch[052] Batch [0749]/[3759]	Speed: 70.135739 samples/sec	 accuracy=65.464583	 loss=1.402497	 lr=0.001000
Epoch[052] Batch [0799]/[3759]	Speed: 69.733973 samples/sec	 accuracy=65.421875	 loss=1.405118	 lr=0.001000
Epoch[052] Batch [0849]/[3759]	Speed: 69.611445 samples/sec	 accuracy=65.503676	 loss=1.404300	 lr=0.001000
Epoch[052] Batch [0899]/[3759]	Speed: 69.807682 samples/sec	 accuracy=65.534722	 loss=1.402476	 lr=0.001000
Epoch[052] Batch [0949]/[3759]	Speed: 70.704135 samples/sec	 accuracy=65.476974	 loss=1.404520	 lr=0.001000
Epoch[052] Batch [0999]/[3759]	Speed: 69.565943 samples/sec	 accuracy=65.429688	 loss=1.407182	 lr=0.001000
Epoch[052] Batch [1049]/[3759]	Speed: 70.548695 samples/sec	 accuracy=65.422619	 loss=1.406886	 lr=0.001000
Epoch[052] Batch [1099]/[3759]	Speed: 69.754613 samples/sec	 accuracy=65.481534	 loss=1.406128	 lr=0.001000
Epoch[052] Batch [1149]/[3759]	Speed: 69.302505 samples/sec	 accuracy=65.483696	 loss=1.407012	 lr=0.001000
Epoch[052] Batch [1199]/[3759]	Speed: 69.548601 samples/sec	 accuracy=65.455729	 loss=1.407658	 lr=0.001000
Epoch[052] Batch [1249]/[3759]	Speed: 69.189147 samples/sec	 accuracy=65.460000	 loss=1.406897	 lr=0.001000
Epoch[052] Batch [1299]/[3759]	Speed: 70.782291 samples/sec	 accuracy=65.532452	 loss=1.404664	 lr=0.001000
Epoch[052] Batch [1349]/[3759]	Speed: 66.994174 samples/sec	 accuracy=65.543981	 loss=1.404878	 lr=0.001000
Epoch[052] Batch [1399]/[3759]	Speed: 71.849770 samples/sec	 accuracy=65.534598	 loss=1.405562	 lr=0.001000
Epoch[052] Batch [1449]/[3759]	Speed: 70.280600 samples/sec	 accuracy=65.536638	 loss=1.404814	 lr=0.001000
Epoch[052] Batch [1499]/[3759]	Speed: 69.639514 samples/sec	 accuracy=65.528125	 loss=1.405143	 lr=0.001000
Epoch[052] Batch [1549]/[3759]	Speed: 67.989219 samples/sec	 accuracy=65.490927	 loss=1.407480	 lr=0.001000
Epoch[052] Batch [1599]/[3759]	Speed: 71.450453 samples/sec	 accuracy=65.484375	 loss=1.407773	 lr=0.001000
Epoch[052] Batch [1649]/[3759]	Speed: 69.851819 samples/sec	 accuracy=65.523674	 loss=1.407608	 lr=0.001000
Epoch[052] Batch [1699]/[3759]	Speed: 69.143994 samples/sec	 accuracy=65.550551	 loss=1.406922	 lr=0.001000
Epoch[052] Batch [1749]/[3759]	Speed: 70.230270 samples/sec	 accuracy=65.565179	 loss=1.407111	 lr=0.001000
Epoch[052] Batch [1799]/[3759]	Speed: 69.022692 samples/sec	 accuracy=65.596354	 loss=1.406199	 lr=0.001000
Epoch[052] Batch [1849]/[3759]	Speed: 70.038348 samples/sec	 accuracy=65.593750	 loss=1.405737	 lr=0.001000
Epoch[052] Batch [1899]/[3759]	Speed: 69.232895 samples/sec	 accuracy=65.608553	 loss=1.404744	 lr=0.001000
Epoch[052] Batch [1949]/[3759]	Speed: 70.674860 samples/sec	 accuracy=65.627404	 loss=1.404703	 lr=0.001000
Epoch[052] Batch [1999]/[3759]	Speed: 69.960397 samples/sec	 accuracy=65.651563	 loss=1.404160	 lr=0.001000
Epoch[052] Batch [2049]/[3759]	Speed: 68.889482 samples/sec	 accuracy=65.682165	 loss=1.402375	 lr=0.001000
Epoch[052] Batch [2099]/[3759]	Speed: 70.913168 samples/sec	 accuracy=65.692708	 loss=1.401617	 lr=0.001000
Epoch[052] Batch [2149]/[3759]	Speed: 69.589904 samples/sec	 accuracy=65.679506	 loss=1.401427	 lr=0.001000
Epoch[052] Batch [2199]/[3759]	Speed: 69.331546 samples/sec	 accuracy=65.671875	 loss=1.402319	 lr=0.001000
Epoch[052] Batch [2249]/[3759]	Speed: 68.993260 samples/sec	 accuracy=65.627778	 loss=1.404058	 lr=0.001000
Epoch[052] Batch [2299]/[3759]	Speed: 71.472597 samples/sec	 accuracy=65.619565	 loss=1.404490	 lr=0.001000
Epoch[052] Batch [2349]/[3759]	Speed: 69.329043 samples/sec	 accuracy=65.607048	 loss=1.405170	 lr=0.001000
Epoch[052] Batch [2399]/[3759]	Speed: 69.141345 samples/sec	 accuracy=65.620443	 loss=1.404571	 lr=0.001000
Epoch[052] Batch [2449]/[3759]	Speed: 69.894813 samples/sec	 accuracy=65.625000	 loss=1.404666	 lr=0.001000
Epoch[052] Batch [2499]/[3759]	Speed: 70.935866 samples/sec	 accuracy=65.628125	 loss=1.404837	 lr=0.001000
Epoch[052] Batch [2549]/[3759]	Speed: 69.564318 samples/sec	 accuracy=65.610294	 loss=1.405044	 lr=0.001000
Epoch[052] Batch [2599]/[3759]	Speed: 69.276205 samples/sec	 accuracy=65.600962	 loss=1.405893	 lr=0.001000
Epoch[052] Batch [2649]/[3759]	Speed: 67.612683 samples/sec	 accuracy=65.580189	 loss=1.406175	 lr=0.001000
Epoch[052] Batch [2699]/[3759]	Speed: 71.809909 samples/sec	 accuracy=65.582755	 loss=1.406652	 lr=0.001000
Epoch[052] Batch [2749]/[3759]	Speed: 70.314909 samples/sec	 accuracy=65.593750	 loss=1.406520	 lr=0.001000
Epoch[052] Batch [2799]/[3759]	Speed: 69.721211 samples/sec	 accuracy=65.618862	 loss=1.405402	 lr=0.001000
Epoch[052] Batch [2849]/[3759]	Speed: 68.848566 samples/sec	 accuracy=65.628289	 loss=1.405818	 lr=0.001000
Epoch[052] Batch [2899]/[3759]	Speed: 70.348175 samples/sec	 accuracy=65.633621	 loss=1.405751	 lr=0.001000
Epoch[052] Batch [2949]/[3759]	Speed: 67.991366 samples/sec	 accuracy=65.600636	 loss=1.407105	 lr=0.001000
Epoch[052] Batch [2999]/[3759]	Speed: 71.199185 samples/sec	 accuracy=65.600000	 loss=1.407385	 lr=0.001000
Epoch[052] Batch [3049]/[3759]	Speed: 69.051805 samples/sec	 accuracy=65.618340	 loss=1.406920	 lr=0.001000
Epoch[052] Batch [3099]/[3759]	Speed: 68.737936 samples/sec	 accuracy=65.609879	 loss=1.407384	 lr=0.001000
Epoch[052] Batch [3149]/[3759]	Speed: 69.917854 samples/sec	 accuracy=65.602679	 loss=1.408434	 lr=0.001000
Epoch[052] Batch [3199]/[3759]	Speed: 71.198732 samples/sec	 accuracy=65.612305	 loss=1.407958	 lr=0.001000
Epoch[052] Batch [3249]/[3759]	Speed: 69.499574 samples/sec	 accuracy=65.626923	 loss=1.407618	 lr=0.001000
Epoch[052] Batch [3299]/[3759]	Speed: 68.317126 samples/sec	 accuracy=65.621686	 loss=1.407586	 lr=0.001000
Epoch[052] Batch [3349]/[3759]	Speed: 70.199456 samples/sec	 accuracy=65.597481	 loss=1.408462	 lr=0.001000
Epoch[052] Batch [3399]/[3759]	Speed: 70.252250 samples/sec	 accuracy=65.598346	 loss=1.408716	 lr=0.001000
Epoch[052] Batch [3449]/[3759]	Speed: 68.270683 samples/sec	 accuracy=65.586957	 loss=1.409309	 lr=0.001000
Epoch[052] Batch [3499]/[3759]	Speed: 71.220426 samples/sec	 accuracy=65.589286	 loss=1.408881	 lr=0.001000
Epoch[052] Batch [3549]/[3759]	Speed: 71.246143 samples/sec	 accuracy=65.559419	 loss=1.409911	 lr=0.001000
Epoch[052] Batch [3599]/[3759]	Speed: 67.925920 samples/sec	 accuracy=65.551215	 loss=1.410572	 lr=0.001000
Epoch[052] Batch [3649]/[3759]	Speed: 70.828242 samples/sec	 accuracy=65.563356	 loss=1.410290	 lr=0.001000
Epoch[052] Batch [3699]/[3759]	Speed: 68.695640 samples/sec	 accuracy=65.568412	 loss=1.410273	 lr=0.001000
Epoch[052] Batch [3749]/[3759]	Speed: 77.948838 samples/sec	 accuracy=65.568750	 loss=1.409956	 lr=0.001000
Batch [0049]/[0303]: acc-top1=61.031250 acc-top5=83.781250
Batch [0099]/[0303]: acc-top1=61.640625 acc-top5=83.609375
Batch [0149]/[0303]: acc-top1=62.708333 acc-top5=83.822917
Batch [0199]/[0303]: acc-top1=62.898438 acc-top5=84.203125
Batch [0249]/[0303]: acc-top1=62.912500 acc-top5=84.362500
Batch [0299]/[0303]: acc-top1=62.697917 acc-top5=84.177083
[Epoch 052] training: accuracy=65.572210	 loss=1.410007
[Epoch 052] speed: 69 samples/sec	time cost: 3751.173542
[Epoch 052] validation: acc-top1=62.665017 acc-top5=84.179043 loss=1.616696
Epoch[053] Batch [0049]/[3760]	Speed: 46.885691 samples/sec	 accuracy=66.500000	 loss=1.394891	 lr=0.001000
Epoch[053] Batch [0099]/[3760]	Speed: 69.511304 samples/sec	 accuracy=66.359375	 loss=1.391383	 lr=0.001000
Epoch[053] Batch [0149]/[3760]	Speed: 70.283965 samples/sec	 accuracy=66.520833	 loss=1.382689	 lr=0.001000
Epoch[053] Batch [0199]/[3760]	Speed: 70.275600 samples/sec	 accuracy=66.492188	 loss=1.382009	 lr=0.001000
Epoch[053] Batch [0249]/[3760]	Speed: 69.980967 samples/sec	 accuracy=66.462500	 loss=1.386198	 lr=0.001000
Epoch[053] Batch [0299]/[3760]	Speed: 69.157625 samples/sec	 accuracy=66.244792	 loss=1.393576	 lr=0.001000
Epoch[053] Batch [0349]/[3760]	Speed: 67.673679 samples/sec	 accuracy=66.133929	 loss=1.392287	 lr=0.001000
Epoch[053] Batch [0399]/[3760]	Speed: 71.070967 samples/sec	 accuracy=66.175781	 loss=1.389671	 lr=0.001000
Epoch[053] Batch [0449]/[3760]	Speed: 70.385712 samples/sec	 accuracy=66.142361	 loss=1.393422	 lr=0.001000
Epoch[053] Batch [0499]/[3760]	Speed: 68.824742 samples/sec	 accuracy=66.018750	 loss=1.395269	 lr=0.001000
Epoch[053] Batch [0549]/[3760]	Speed: 70.096028 samples/sec	 accuracy=66.042614	 loss=1.395256	 lr=0.001000
Epoch[053] Batch [0599]/[3760]	Speed: 68.659141 samples/sec	 accuracy=66.005208	 loss=1.395415	 lr=0.001000
Epoch[053] Batch [0649]/[3760]	Speed: 70.928315 samples/sec	 accuracy=65.951923	 loss=1.396528	 lr=0.001000
Epoch[053] Batch [0699]/[3760]	Speed: 69.553424 samples/sec	 accuracy=65.908482	 loss=1.399918	 lr=0.001000
Epoch[053] Batch [0749]/[3760]	Speed: 69.342874 samples/sec	 accuracy=65.972917	 loss=1.395871	 lr=0.001000
Epoch[053] Batch [0799]/[3760]	Speed: 69.811292 samples/sec	 accuracy=65.984375	 loss=1.394295	 lr=0.001000
Epoch[053] Batch [0849]/[3760]	Speed: 69.761583 samples/sec	 accuracy=65.998162	 loss=1.394263	 lr=0.001000
Epoch[053] Batch [0899]/[3760]	Speed: 69.407997 samples/sec	 accuracy=66.008681	 loss=1.392424	 lr=0.001000
Epoch[053] Batch [0949]/[3760]	Speed: 66.670182 samples/sec	 accuracy=66.080592	 loss=1.390433	 lr=0.001000
Epoch[053] Batch [0999]/[3760]	Speed: 72.385880 samples/sec	 accuracy=66.053125	 loss=1.393916	 lr=0.001000
Epoch[053] Batch [1049]/[3760]	Speed: 67.757805 samples/sec	 accuracy=66.078869	 loss=1.392581	 lr=0.001000
Epoch[053] Batch [1099]/[3760]	Speed: 69.687164 samples/sec	 accuracy=66.078125	 loss=1.394342	 lr=0.001000
Epoch[053] Batch [1149]/[3760]	Speed: 70.869061 samples/sec	 accuracy=66.020380	 loss=1.395502	 lr=0.001000
Epoch[053] Batch [1199]/[3760]	Speed: 70.574467 samples/sec	 accuracy=66.062500	 loss=1.392690	 lr=0.001000
Epoch[053] Batch [1249]/[3760]	Speed: 69.716493 samples/sec	 accuracy=66.047500	 loss=1.392496	 lr=0.001000
Epoch[053] Batch [1299]/[3760]	Speed: 69.424704 samples/sec	 accuracy=66.016827	 loss=1.393369	 lr=0.001000
Epoch[053] Batch [1349]/[3760]	Speed: 69.651834 samples/sec	 accuracy=65.993056	 loss=1.393836	 lr=0.001000
Epoch[053] Batch [1399]/[3760]	Speed: 69.709012 samples/sec	 accuracy=66.044643	 loss=1.392730	 lr=0.001000
Epoch[053] Batch [1449]/[3760]	Speed: 68.513101 samples/sec	 accuracy=66.052802	 loss=1.392657	 lr=0.001000
Epoch[053] Batch [1499]/[3760]	Speed: 69.200534 samples/sec	 accuracy=65.998958	 loss=1.393938	 lr=0.001000
Epoch[053] Batch [1549]/[3760]	Speed: 67.744958 samples/sec	 accuracy=65.987903	 loss=1.393534	 lr=0.001000
Epoch[053] Batch [1599]/[3760]	Speed: 71.740647 samples/sec	 accuracy=65.962891	 loss=1.394375	 lr=0.001000
Epoch[053] Batch [1649]/[3760]	Speed: 70.865695 samples/sec	 accuracy=65.971591	 loss=1.393658	 lr=0.001000
Epoch[053] Batch [1699]/[3760]	Speed: 68.795911 samples/sec	 accuracy=65.951287	 loss=1.393111	 lr=0.001000
Epoch[053] Batch [1749]/[3760]	Speed: 70.223712 samples/sec	 accuracy=65.924107	 loss=1.393718	 lr=0.001000
Epoch[053] Batch [1799]/[3760]	Speed: 69.397836 samples/sec	 accuracy=65.921875	 loss=1.394116	 lr=0.001000
Epoch[053] Batch [1849]/[3760]	Speed: 68.629772 samples/sec	 accuracy=65.927365	 loss=1.393719	 lr=0.001000
Epoch[053] Batch [1899]/[3760]	Speed: 71.348360 samples/sec	 accuracy=65.917763	 loss=1.394171	 lr=0.001000
Epoch[053] Batch [1949]/[3760]	Speed: 69.206897 samples/sec	 accuracy=65.944712	 loss=1.393659	 lr=0.001000
Epoch[053] Batch [1999]/[3760]	Speed: 67.542186 samples/sec	 accuracy=65.949219	 loss=1.392633	 lr=0.001000
Epoch[053] Batch [2049]/[3760]	Speed: 70.206170 samples/sec	 accuracy=65.931402	 loss=1.394148	 lr=0.001000
Epoch[053] Batch [2099]/[3760]	Speed: 71.799106 samples/sec	 accuracy=65.920387	 loss=1.394842	 lr=0.001000
Epoch[053] Batch [2149]/[3760]	Speed: 68.228504 samples/sec	 accuracy=65.909884	 loss=1.395578	 lr=0.001000
Epoch[053] Batch [2199]/[3760]	Speed: 70.180005 samples/sec	 accuracy=65.907670	 loss=1.396054	 lr=0.001000
Epoch[053] Batch [2249]/[3760]	Speed: 69.956078 samples/sec	 accuracy=65.890278	 loss=1.396567	 lr=0.001000
Epoch[053] Batch [2299]/[3760]	Speed: 68.196009 samples/sec	 accuracy=65.879755	 loss=1.396851	 lr=0.001000
Epoch[053] Batch [2349]/[3760]	Speed: 71.183178 samples/sec	 accuracy=65.880984	 loss=1.397096	 lr=0.001000
Epoch[053] Batch [2399]/[3760]	Speed: 68.205445 samples/sec	 accuracy=65.878906	 loss=1.397012	 lr=0.001000
Epoch[053] Batch [2449]/[3760]	Speed: 71.206605 samples/sec	 accuracy=65.871173	 loss=1.396998	 lr=0.001000
Epoch[053] Batch [2499]/[3760]	Speed: 69.307030 samples/sec	 accuracy=65.873750	 loss=1.397067	 lr=0.001000
Epoch[053] Batch [2549]/[3760]	Speed: 69.353031 samples/sec	 accuracy=65.862132	 loss=1.397848	 lr=0.001000
Epoch[053] Batch [2599]/[3760]	Speed: 70.048701 samples/sec	 accuracy=65.858774	 loss=1.398489	 lr=0.001000
Epoch[053] Batch [2649]/[3760]	Speed: 69.528653 samples/sec	 accuracy=65.863208	 loss=1.399006	 lr=0.001000
Epoch[053] Batch [2699]/[3760]	Speed: 70.415356 samples/sec	 accuracy=65.836227	 loss=1.399401	 lr=0.001000
Epoch[053] Batch [2749]/[3760]	Speed: 69.821652 samples/sec	 accuracy=65.850000	 loss=1.399493	 lr=0.001000
Epoch[053] Batch [2799]/[3760]	Speed: 66.605523 samples/sec	 accuracy=65.849330	 loss=1.399871	 lr=0.001000
Epoch[053] Batch [2849]/[3760]	Speed: 69.919764 samples/sec	 accuracy=65.832237	 loss=1.400008	 lr=0.001000
Epoch[053] Batch [2899]/[3760]	Speed: 71.714331 samples/sec	 accuracy=65.826509	 loss=1.400205	 lr=0.001000
Epoch[053] Batch [2949]/[3760]	Speed: 70.752305 samples/sec	 accuracy=65.806144	 loss=1.400835	 lr=0.001000
Epoch[053] Batch [2999]/[3760]	Speed: 69.837269 samples/sec	 accuracy=65.834375	 loss=1.400031	 lr=0.001000
Epoch[053] Batch [3049]/[3760]	Speed: 69.165186 samples/sec	 accuracy=65.839652	 loss=1.400137	 lr=0.001000
Epoch[053] Batch [3099]/[3760]	Speed: 69.503026 samples/sec	 accuracy=65.821573	 loss=1.399964	 lr=0.001000
Epoch[053] Batch [3149]/[3760]	Speed: 70.220293 samples/sec	 accuracy=65.815476	 loss=1.399860	 lr=0.001000
Epoch[053] Batch [3199]/[3760]	Speed: 69.045839 samples/sec	 accuracy=65.804688	 loss=1.401243	 lr=0.001000
Epoch[053] Batch [3249]/[3760]	Speed: 68.598577 samples/sec	 accuracy=65.786538	 loss=1.401215	 lr=0.001000
Epoch[053] Batch [3299]/[3760]	Speed: 70.674880 samples/sec	 accuracy=65.800189	 loss=1.400965	 lr=0.001000
Epoch[053] Batch [3349]/[3760]	Speed: 67.803375 samples/sec	 accuracy=65.793377	 loss=1.401384	 lr=0.001000
Epoch[053] Batch [3399]/[3760]	Speed: 70.976419 samples/sec	 accuracy=65.795037	 loss=1.401179	 lr=0.001000
Epoch[053] Batch [3449]/[3760]	Speed: 67.476770 samples/sec	 accuracy=65.768116	 loss=1.402042	 lr=0.001000
Epoch[053] Batch [3499]/[3760]	Speed: 70.573905 samples/sec	 accuracy=65.773214	 loss=1.402037	 lr=0.001000
Epoch[053] Batch [3549]/[3760]	Speed: 70.521645 samples/sec	 accuracy=65.768926	 loss=1.402045	 lr=0.001000
Epoch[053] Batch [3599]/[3760]	Speed: 69.674596 samples/sec	 accuracy=65.764757	 loss=1.402384	 lr=0.001000
Epoch[053] Batch [3649]/[3760]	Speed: 69.797711 samples/sec	 accuracy=65.784675	 loss=1.401858	 lr=0.001000
Epoch[053] Batch [3699]/[3760]	Speed: 70.399274 samples/sec	 accuracy=65.782095	 loss=1.401979	 lr=0.001000
Epoch[053] Batch [3749]/[3760]	Speed: 75.964764 samples/sec	 accuracy=65.776667	 loss=1.402475	 lr=0.001000
Batch [0049]/[0303]: acc-top1=61.218750 acc-top5=83.812500
Batch [0099]/[0303]: acc-top1=61.171875 acc-top5=83.500000
Batch [0149]/[0303]: acc-top1=61.958333 acc-top5=83.791667
Batch [0199]/[0303]: acc-top1=62.359375 acc-top5=84.039062
Batch [0249]/[0303]: acc-top1=62.562500 acc-top5=84.287500
Batch [0299]/[0303]: acc-top1=62.375000 acc-top5=84.125000
[Epoch 053] training: accuracy=65.776679	 loss=1.402705
[Epoch 053] speed: 69 samples/sec	time cost: 3757.862211
[Epoch 053] validation: acc-top1=62.355611 acc-top5=84.142946 loss=1.616152
Epoch[054] Batch [0049]/[3760]	Speed: 48.599260 samples/sec	 accuracy=67.437500	 loss=1.344593	 lr=0.001000
Epoch[054] Batch [0099]/[3760]	Speed: 67.580826 samples/sec	 accuracy=66.343750	 loss=1.385642	 lr=0.001000
Epoch[054] Batch [0149]/[3760]	Speed: 69.553993 samples/sec	 accuracy=66.239583	 loss=1.388854	 lr=0.001000
Epoch[054] Batch [0199]/[3760]	Speed: 69.122662 samples/sec	 accuracy=66.078125	 loss=1.394061	 lr=0.001000
Epoch[054] Batch [0249]/[3760]	Speed: 71.074661 samples/sec	 accuracy=66.118750	 loss=1.396781	 lr=0.001000
Epoch[054] Batch [0299]/[3760]	Speed: 69.074767 samples/sec	 accuracy=66.026042	 loss=1.399733	 lr=0.001000
Epoch[054] Batch [0349]/[3760]	Speed: 69.199387 samples/sec	 accuracy=65.995536	 loss=1.395813	 lr=0.001000
Epoch[054] Batch [0399]/[3760]	Speed: 70.516256 samples/sec	 accuracy=65.972656	 loss=1.393463	 lr=0.001000
Epoch[054] Batch [0449]/[3760]	Speed: 70.949397 samples/sec	 accuracy=66.048611	 loss=1.390544	 lr=0.001000
Epoch[054] Batch [0499]/[3760]	Speed: 69.886068 samples/sec	 accuracy=66.078125	 loss=1.387803	 lr=0.001000
Epoch[054] Batch [0549]/[3760]	Speed: 69.022974 samples/sec	 accuracy=65.974432	 loss=1.388032	 lr=0.001000
Epoch[054] Batch [0599]/[3760]	Speed: 68.395856 samples/sec	 accuracy=65.825521	 loss=1.394972	 lr=0.001000
Epoch[054] Batch [0649]/[3760]	Speed: 68.258811 samples/sec	 accuracy=65.935096	 loss=1.392145	 lr=0.001000
Epoch[054] Batch [0699]/[3760]	Speed: 71.769119 samples/sec	 accuracy=65.966518	 loss=1.389563	 lr=0.001000
Epoch[054] Batch [0749]/[3760]	Speed: 68.277268 samples/sec	 accuracy=65.910417	 loss=1.392286	 lr=0.001000
Epoch[054] Batch [0799]/[3760]	Speed: 71.170266 samples/sec	 accuracy=65.974609	 loss=1.389833	 lr=0.001000
Epoch[054] Batch [0849]/[3760]	Speed: 68.835807 samples/sec	 accuracy=66.011029	 loss=1.390504	 lr=0.001000
Epoch[054] Batch [0899]/[3760]	Speed: 69.975554 samples/sec	 accuracy=66.027778	 loss=1.388739	 lr=0.001000
Epoch[054] Batch [0949]/[3760]	Speed: 69.508615 samples/sec	 accuracy=66.036184	 loss=1.388259	 lr=0.001000
Epoch[054] Batch [0999]/[3760]	Speed: 69.420265 samples/sec	 accuracy=66.032813	 loss=1.388463	 lr=0.001000
Epoch[054] Batch [1049]/[3760]	Speed: 66.646518 samples/sec	 accuracy=66.046131	 loss=1.387451	 lr=0.001000
Epoch[054] Batch [1099]/[3760]	Speed: 73.307518 samples/sec	 accuracy=66.019886	 loss=1.388570	 lr=0.001000
Epoch[054] Batch [1149]/[3760]	Speed: 65.687916 samples/sec	 accuracy=65.975543	 loss=1.390236	 lr=0.001000
Epoch[054] Batch [1199]/[3760]	Speed: 74.011260 samples/sec	 accuracy=65.950521	 loss=1.391213	 lr=0.001000
Epoch[054] Batch [1249]/[3760]	Speed: 68.657126 samples/sec	 accuracy=65.912500	 loss=1.391896	 lr=0.001000
Epoch[054] Batch [1299]/[3760]	Speed: 70.076766 samples/sec	 accuracy=65.894231	 loss=1.392807	 lr=0.001000
Epoch[054] Batch [1349]/[3760]	Speed: 70.790455 samples/sec	 accuracy=65.856481	 loss=1.393802	 lr=0.001000
Epoch[054] Batch [1399]/[3760]	Speed: 68.228780 samples/sec	 accuracy=65.796875	 loss=1.395950	 lr=0.001000
Epoch[054] Batch [1449]/[3760]	Speed: 70.953753 samples/sec	 accuracy=65.759698	 loss=1.396762	 lr=0.001000
Epoch[054] Batch [1499]/[3760]	Speed: 70.006675 samples/sec	 accuracy=65.776042	 loss=1.397445	 lr=0.001000
Epoch[054] Batch [1549]/[3760]	Speed: 69.955928 samples/sec	 accuracy=65.752016	 loss=1.398282	 lr=0.001000
Epoch[054] Batch [1599]/[3760]	Speed: 68.337737 samples/sec	 accuracy=65.742188	 loss=1.399108	 lr=0.001000
Epoch[054] Batch [1649]/[3760]	Speed: 71.549221 samples/sec	 accuracy=65.721591	 loss=1.400015	 lr=0.001000
Epoch[054] Batch [1699]/[3760]	Speed: 70.223832 samples/sec	 accuracy=65.740809	 loss=1.399287	 lr=0.001000
Epoch[054] Batch [1749]/[3760]	Speed: 69.836936 samples/sec	 accuracy=65.706250	 loss=1.400014	 lr=0.001000
Epoch[054] Batch [1799]/[3760]	Speed: 68.613783 samples/sec	 accuracy=65.729167	 loss=1.399385	 lr=0.001000
Epoch[054] Batch [1849]/[3760]	Speed: 68.768706 samples/sec	 accuracy=65.754223	 loss=1.398460	 lr=0.001000
Epoch[054] Batch [1899]/[3760]	Speed: 71.470274 samples/sec	 accuracy=65.707237	 loss=1.399907	 lr=0.001000
Epoch[054] Batch [1949]/[3760]	Speed: 68.547224 samples/sec	 accuracy=65.710737	 loss=1.399252	 lr=0.001000
Epoch[054] Batch [1999]/[3760]	Speed: 70.131105 samples/sec	 accuracy=65.697656	 loss=1.400778	 lr=0.001000
Epoch[054] Batch [2049]/[3760]	Speed: 69.687060 samples/sec	 accuracy=65.686738	 loss=1.400326	 lr=0.001000
Epoch[054] Batch [2099]/[3760]	Speed: 69.027360 samples/sec	 accuracy=65.691964	 loss=1.400440	 lr=0.001000
Epoch[054] Batch [2149]/[3760]	Speed: 69.784256 samples/sec	 accuracy=65.694041	 loss=1.400101	 lr=0.001000
Epoch[054] Batch [2199]/[3760]	Speed: 69.786402 samples/sec	 accuracy=65.676136	 loss=1.399945	 lr=0.001000
Epoch[054] Batch [2249]/[3760]	Speed: 70.194592 samples/sec	 accuracy=65.652083	 loss=1.401031	 lr=0.001000
Epoch[054] Batch [2299]/[3760]	Speed: 68.220864 samples/sec	 accuracy=65.679348	 loss=1.400024	 lr=0.001000
Epoch[054] Batch [2349]/[3760]	Speed: 71.143402 samples/sec	 accuracy=65.661569	 loss=1.400719	 lr=0.001000
Epoch[054] Batch [2399]/[3760]	Speed: 70.127660 samples/sec	 accuracy=65.630208	 loss=1.401708	 lr=0.001000
Epoch[054] Batch [2449]/[3760]	Speed: 69.315170 samples/sec	 accuracy=65.601403	 loss=1.402325	 lr=0.001000
Epoch[054] Batch [2499]/[3760]	Speed: 70.161928 samples/sec	 accuracy=65.580625	 loss=1.402691	 lr=0.001000
Epoch[054] Batch [2549]/[3760]	Speed: 68.998774 samples/sec	 accuracy=65.586397	 loss=1.402707	 lr=0.001000
Epoch[054] Batch [2599]/[3760]	Speed: 70.495863 samples/sec	 accuracy=65.602163	 loss=1.402652	 lr=0.001000
Epoch[054] Batch [2649]/[3760]	Speed: 70.631901 samples/sec	 accuracy=65.586085	 loss=1.403494	 lr=0.001000
Epoch[054] Batch [2699]/[3760]	Speed: 69.329832 samples/sec	 accuracy=65.600116	 loss=1.403206	 lr=0.001000
Epoch[054] Batch [2749]/[3760]	Speed: 69.127509 samples/sec	 accuracy=65.604545	 loss=1.403004	 lr=0.001000
Epoch[054] Batch [2799]/[3760]	Speed: 69.907203 samples/sec	 accuracy=65.573661	 loss=1.403994	 lr=0.001000
Epoch[054] Batch [2849]/[3760]	Speed: 70.838831 samples/sec	 accuracy=65.576754	 loss=1.403932	 lr=0.001000
Epoch[054] Batch [2899]/[3760]	Speed: 69.449651 samples/sec	 accuracy=65.582435	 loss=1.403538	 lr=0.001000
Epoch[054] Batch [2949]/[3760]	Speed: 68.859651 samples/sec	 accuracy=65.581038	 loss=1.403583	 lr=0.001000
Epoch[054] Batch [2999]/[3760]	Speed: 70.203970 samples/sec	 accuracy=65.608854	 loss=1.402116	 lr=0.001000
Epoch[054] Batch [3049]/[3760]	Speed: 70.271411 samples/sec	 accuracy=65.597336	 loss=1.402063	 lr=0.001000
Epoch[054] Batch [3099]/[3760]	Speed: 70.371603 samples/sec	 accuracy=65.578629	 loss=1.402501	 lr=0.001000
Epoch[054] Batch [3149]/[3760]	Speed: 69.957251 samples/sec	 accuracy=65.561508	 loss=1.403022	 lr=0.001000
Epoch[054] Batch [3199]/[3760]	Speed: 68.760924 samples/sec	 accuracy=65.574707	 loss=1.402777	 lr=0.001000
Epoch[054] Batch [3249]/[3760]	Speed: 68.743955 samples/sec	 accuracy=65.587981	 loss=1.402370	 lr=0.001000
Epoch[054] Batch [3299]/[3760]	Speed: 69.306108 samples/sec	 accuracy=65.572917	 loss=1.402734	 lr=0.001000
Epoch[054] Batch [3349]/[3760]	Speed: 70.114815 samples/sec	 accuracy=65.581623	 loss=1.402571	 lr=0.001000
Epoch[054] Batch [3399]/[3760]	Speed: 70.331258 samples/sec	 accuracy=65.583180	 loss=1.402697	 lr=0.001000
Epoch[054] Batch [3449]/[3760]	Speed: 69.732984 samples/sec	 accuracy=65.574275	 loss=1.402670	 lr=0.001000
Epoch[054] Batch [3499]/[3760]	Speed: 71.000829 samples/sec	 accuracy=65.579018	 loss=1.402545	 lr=0.001000
Epoch[054] Batch [3549]/[3760]	Speed: 70.100553 samples/sec	 accuracy=65.569102	 loss=1.402947	 lr=0.001000
Epoch[054] Batch [3599]/[3760]	Speed: 68.647360 samples/sec	 accuracy=65.578993	 loss=1.402865	 lr=0.001000
Epoch[054] Batch [3649]/[3760]	Speed: 70.727141 samples/sec	 accuracy=65.575342	 loss=1.403079	 lr=0.001000
Epoch[054] Batch [3699]/[3760]	Speed: 69.007147 samples/sec	 accuracy=65.578970	 loss=1.402804	 lr=0.001000
Epoch[054] Batch [3749]/[3760]	Speed: 76.173776 samples/sec	 accuracy=65.591250	 loss=1.402214	 lr=0.001000
Batch [0049]/[0303]: acc-top1=61.531250 acc-top5=83.968750
Batch [0099]/[0303]: acc-top1=61.812500 acc-top5=83.625000
Batch [0149]/[0303]: acc-top1=62.583333 acc-top5=83.958333
Batch [0199]/[0303]: acc-top1=62.960938 acc-top5=84.273438
Batch [0249]/[0303]: acc-top1=63.093750 acc-top5=84.462500
Batch [0299]/[0303]: acc-top1=62.781250 acc-top5=84.281250
[Epoch 054] training: accuracy=65.586353	 loss=1.402429
[Epoch 054] speed: 69 samples/sec	time cost: 3752.978800
[Epoch 054] validation: acc-top1=62.783622 acc-top5=84.287335 loss=1.611367
Epoch[055] Batch [0049]/[3759]	Speed: 48.704505 samples/sec	 accuracy=66.031250	 loss=1.387979	 lr=0.001000
Epoch[055] Batch [0099]/[3759]	Speed: 69.396771 samples/sec	 accuracy=66.328125	 loss=1.376967	 lr=0.001000
Epoch[055] Batch [0149]/[3759]	Speed: 68.431642 samples/sec	 accuracy=66.385417	 loss=1.376832	 lr=0.001000
Epoch[055] Batch [0199]/[3759]	Speed: 69.357594 samples/sec	 accuracy=66.210938	 loss=1.378025	 lr=0.001000
Epoch[055] Batch [0249]/[3759]	Speed: 70.095030 samples/sec	 accuracy=66.250000	 loss=1.378758	 lr=0.001000
Epoch[055] Batch [0299]/[3759]	Speed: 70.307546 samples/sec	 accuracy=66.223958	 loss=1.383245	 lr=0.001000
Epoch[055] Batch [0349]/[3759]	Speed: 69.653761 samples/sec	 accuracy=66.058036	 loss=1.386717	 lr=0.001000
Epoch[055] Batch [0399]/[3759]	Speed: 70.541193 samples/sec	 accuracy=66.132812	 loss=1.389983	 lr=0.001000
Epoch[055] Batch [0449]/[3759]	Speed: 69.720412 samples/sec	 accuracy=66.232639	 loss=1.385734	 lr=0.001000
Epoch[055] Batch [0499]/[3759]	Speed: 70.476498 samples/sec	 accuracy=66.150000	 loss=1.387554	 lr=0.001000
Epoch[055] Batch [0549]/[3759]	Speed: 68.173548 samples/sec	 accuracy=66.031250	 loss=1.389733	 lr=0.001000
Epoch[055] Batch [0599]/[3759]	Speed: 70.208384 samples/sec	 accuracy=66.088542	 loss=1.388603	 lr=0.001000
Epoch[055] Batch [0649]/[3759]	Speed: 70.281018 samples/sec	 accuracy=66.120192	 loss=1.389988	 lr=0.001000
Epoch[055] Batch [0699]/[3759]	Speed: 70.141415 samples/sec	 accuracy=66.158482	 loss=1.389114	 lr=0.001000
Epoch[055] Batch [0749]/[3759]	Speed: 69.230573 samples/sec	 accuracy=66.147917	 loss=1.387721	 lr=0.001000
Epoch[055] Batch [0799]/[3759]	Speed: 69.570707 samples/sec	 accuracy=66.029297	 loss=1.388128	 lr=0.001000
Epoch[055] Batch [0849]/[3759]	Speed: 70.374097 samples/sec	 accuracy=65.966912	 loss=1.391032	 lr=0.001000
Epoch[055] Batch [0899]/[3759]	Speed: 67.655010 samples/sec	 accuracy=66.020833	 loss=1.391083	 lr=0.001000
Epoch[055] Batch [0949]/[3759]	Speed: 71.859205 samples/sec	 accuracy=66.044408	 loss=1.391422	 lr=0.001000
Epoch[055] Batch [0999]/[3759]	Speed: 70.493035 samples/sec	 accuracy=66.037500	 loss=1.393588	 lr=0.001000
Epoch[055] Batch [1049]/[3759]	Speed: 68.410980 samples/sec	 accuracy=66.004464	 loss=1.395278	 lr=0.001000
Epoch[055] Batch [1099]/[3759]	Speed: 70.712911 samples/sec	 accuracy=65.992898	 loss=1.395510	 lr=0.001000
Epoch[055] Batch [1149]/[3759]	Speed: 67.849897 samples/sec	 accuracy=65.925272	 loss=1.397805	 lr=0.001000
Epoch[055] Batch [1199]/[3759]	Speed: 71.412244 samples/sec	 accuracy=65.876302	 loss=1.397573	 lr=0.001000
Epoch[055] Batch [1249]/[3759]	Speed: 68.122855 samples/sec	 accuracy=65.898750	 loss=1.396440	 lr=0.001000
Epoch[055] Batch [1299]/[3759]	Speed: 69.570208 samples/sec	 accuracy=65.927885	 loss=1.395181	 lr=0.001000
Epoch[055] Batch [1349]/[3759]	Speed: 69.443023 samples/sec	 accuracy=65.922454	 loss=1.395010	 lr=0.001000
Epoch[055] Batch [1399]/[3759]	Speed: 70.813215 samples/sec	 accuracy=65.959821	 loss=1.394443	 lr=0.001000
Epoch[055] Batch [1449]/[3759]	Speed: 70.173090 samples/sec	 accuracy=65.979526	 loss=1.393932	 lr=0.001000
Epoch[055] Batch [1499]/[3759]	Speed: 67.386356 samples/sec	 accuracy=65.946875	 loss=1.395530	 lr=0.001000
Epoch[055] Batch [1549]/[3759]	Speed: 70.871151 samples/sec	 accuracy=65.933468	 loss=1.395354	 lr=0.001000
Epoch[055] Batch [1599]/[3759]	Speed: 67.500835 samples/sec	 accuracy=65.933594	 loss=1.395264	 lr=0.001000
Epoch[055] Batch [1649]/[3759]	Speed: 69.655265 samples/sec	 accuracy=65.950758	 loss=1.394041	 lr=0.001000
Epoch[055] Batch [1699]/[3759]	Speed: 70.151925 samples/sec	 accuracy=65.959559	 loss=1.393155	 lr=0.001000
Epoch[055] Batch [1749]/[3759]	Speed: 70.033802 samples/sec	 accuracy=65.962500	 loss=1.392118	 lr=0.001000
Epoch[055] Batch [1799]/[3759]	Speed: 69.925085 samples/sec	 accuracy=65.980035	 loss=1.392167	 lr=0.001000
Epoch[055] Batch [1849]/[3759]	Speed: 70.421780 samples/sec	 accuracy=65.989020	 loss=1.391559	 lr=0.001000
Epoch[055] Batch [1899]/[3759]	Speed: 70.508435 samples/sec	 accuracy=65.951480	 loss=1.392733	 lr=0.001000
Epoch[055] Batch [1949]/[3759]	Speed: 70.126086 samples/sec	 accuracy=65.943910	 loss=1.392583	 lr=0.001000
Epoch[055] Batch [1999]/[3759]	Speed: 69.509306 samples/sec	 accuracy=65.940625	 loss=1.392119	 lr=0.001000
Epoch[055] Batch [2049]/[3759]	Speed: 70.531243 samples/sec	 accuracy=65.928354	 loss=1.392666	 lr=0.001000
Epoch[055] Batch [2099]/[3759]	Speed: 69.805439 samples/sec	 accuracy=65.921875	 loss=1.392668	 lr=0.001000
Epoch[055] Batch [2149]/[3759]	Speed: 68.807620 samples/sec	 accuracy=65.917878	 loss=1.393276	 lr=0.001000
Epoch[055] Batch [2199]/[3759]	Speed: 69.762574 samples/sec	 accuracy=65.926847	 loss=1.392956	 lr=0.001000
Epoch[055] Batch [2249]/[3759]	Speed: 68.182951 samples/sec	 accuracy=65.918056	 loss=1.392981	 lr=0.001000
Epoch[055] Batch [2299]/[3759]	Speed: 70.317810 samples/sec	 accuracy=65.919837	 loss=1.392910	 lr=0.001000
Epoch[055] Batch [2349]/[3759]	Speed: 70.137374 samples/sec	 accuracy=65.932846	 loss=1.392159	 lr=0.001000
Epoch[055] Batch [2399]/[3759]	Speed: 69.406824 samples/sec	 accuracy=65.926432	 loss=1.391674	 lr=0.001000
Epoch[055] Batch [2449]/[3759]	Speed: 69.952872 samples/sec	 accuracy=65.903699	 loss=1.392149	 lr=0.001000
Epoch[055] Batch [2499]/[3759]	Speed: 69.388241 samples/sec	 accuracy=65.901250	 loss=1.392472	 lr=0.001000
Epoch[055] Batch [2549]/[3759]	Speed: 68.041611 samples/sec	 accuracy=65.901961	 loss=1.392448	 lr=0.001000
Epoch[055] Batch [2599]/[3759]	Speed: 71.448327 samples/sec	 accuracy=65.898438	 loss=1.392721	 lr=0.001000
Epoch[055] Batch [2649]/[3759]	Speed: 70.162188 samples/sec	 accuracy=65.894458	 loss=1.392932	 lr=0.001000
Epoch[055] Batch [2699]/[3759]	Speed: 69.986419 samples/sec	 accuracy=65.905671	 loss=1.392124	 lr=0.001000
Epoch[055] Batch [2749]/[3759]	Speed: 69.387653 samples/sec	 accuracy=65.896023	 loss=1.392743	 lr=0.001000
Epoch[055] Batch [2799]/[3759]	Speed: 66.842081 samples/sec	 accuracy=65.896763	 loss=1.393018	 lr=0.001000
Epoch[055] Batch [2849]/[3759]	Speed: 71.522157 samples/sec	 accuracy=65.879934	 loss=1.393749	 lr=0.001000
Epoch[055] Batch [2899]/[3759]	Speed: 70.304875 samples/sec	 accuracy=65.869612	 loss=1.394393	 lr=0.001000
Epoch[055] Batch [2949]/[3759]	Speed: 67.574682 samples/sec	 accuracy=65.879237	 loss=1.393873	 lr=0.001000
Epoch[055] Batch [2999]/[3759]	Speed: 72.269416 samples/sec	 accuracy=65.891146	 loss=1.393513	 lr=0.001000
Epoch[055] Batch [3049]/[3759]	Speed: 69.799073 samples/sec	 accuracy=65.901639	 loss=1.392784	 lr=0.001000
Epoch[055] Batch [3099]/[3759]	Speed: 70.392959 samples/sec	 accuracy=65.899698	 loss=1.392808	 lr=0.001000
Epoch[055] Batch [3149]/[3759]	Speed: 68.719070 samples/sec	 accuracy=65.908730	 loss=1.392475	 lr=0.001000
Epoch[055] Batch [3199]/[3759]	Speed: 69.943903 samples/sec	 accuracy=65.899902	 loss=1.393357	 lr=0.001000
Epoch[055] Batch [3249]/[3759]	Speed: 69.896404 samples/sec	 accuracy=65.909615	 loss=1.392654	 lr=0.001000
Epoch[055] Batch [3299]/[3759]	Speed: 69.185911 samples/sec	 accuracy=65.929924	 loss=1.391940	 lr=0.001000
Epoch[055] Batch [3349]/[3759]	Speed: 69.298184 samples/sec	 accuracy=65.928638	 loss=1.392320	 lr=0.001000
Epoch[055] Batch [3399]/[3759]	Speed: 68.952427 samples/sec	 accuracy=65.929228	 loss=1.392350	 lr=0.001000
Epoch[055] Batch [3449]/[3759]	Speed: 69.928927 samples/sec	 accuracy=65.923460	 loss=1.392864	 lr=0.001000
Epoch[055] Batch [3499]/[3759]	Speed: 70.034293 samples/sec	 accuracy=65.920982	 loss=1.393012	 lr=0.001000
Epoch[055] Batch [3549]/[3759]	Speed: 70.072501 samples/sec	 accuracy=65.916813	 loss=1.392368	 lr=0.001000
Epoch[055] Batch [3599]/[3759]	Speed: 68.843832 samples/sec	 accuracy=65.946615	 loss=1.391705	 lr=0.001000
Epoch[055] Batch [3649]/[3759]	Speed: 70.284726 samples/sec	 accuracy=65.945205	 loss=1.391831	 lr=0.001000
Epoch[055] Batch [3699]/[3759]	Speed: 69.023040 samples/sec	 accuracy=65.946791	 loss=1.391680	 lr=0.001000
Epoch[055] Batch [3749]/[3759]	Speed: 77.827357 samples/sec	 accuracy=65.936667	 loss=1.391836	 lr=0.001000
Batch [0049]/[0303]: acc-top1=62.093750 acc-top5=83.968750
Batch [0099]/[0303]: acc-top1=61.656250 acc-top5=83.734375
Batch [0149]/[0303]: acc-top1=62.562500 acc-top5=84.062500
Batch [0199]/[0303]: acc-top1=62.890625 acc-top5=84.421875
Batch [0249]/[0303]: acc-top1=63.056250 acc-top5=84.625000
Batch [0299]/[0303]: acc-top1=62.864583 acc-top5=84.406250
[Epoch 055] training: accuracy=65.934258	 loss=1.391850
[Epoch 055] speed: 69 samples/sec	time cost: 3754.622686
[Epoch 055] validation: acc-top1=62.845503 acc-top5=84.411097 loss=1.605728
Epoch[056] Batch [0049]/[3760]	Speed: 47.779409 samples/sec	 accuracy=65.937500	 loss=1.363471	 lr=0.001000
Epoch[056] Batch [0099]/[3760]	Speed: 69.357322 samples/sec	 accuracy=66.343750	 loss=1.370033	 lr=0.001000
Epoch[056] Batch [0149]/[3760]	Speed: 70.924790 samples/sec	 accuracy=66.635417	 loss=1.368098	 lr=0.001000
Epoch[056] Batch [0199]/[3760]	Speed: 69.790372 samples/sec	 accuracy=66.515625	 loss=1.369706	 lr=0.001000
Epoch[056] Batch [0249]/[3760]	Speed: 69.778047 samples/sec	 accuracy=66.275000	 loss=1.374831	 lr=0.001000
Epoch[056] Batch [0299]/[3760]	Speed: 69.986250 samples/sec	 accuracy=66.109375	 loss=1.376549	 lr=0.001000
Epoch[056] Batch [0349]/[3760]	Speed: 66.627194 samples/sec	 accuracy=66.281250	 loss=1.372168	 lr=0.001000
Epoch[056] Batch [0399]/[3760]	Speed: 68.980785 samples/sec	 accuracy=66.289062	 loss=1.371961	 lr=0.001000
Epoch[056] Batch [0449]/[3760]	Speed: 72.624124 samples/sec	 accuracy=66.204861	 loss=1.375883	 lr=0.001000
Epoch[056] Batch [0499]/[3760]	Speed: 68.371773 samples/sec	 accuracy=66.065625	 loss=1.383192	 lr=0.001000
Epoch[056] Batch [0549]/[3760]	Speed: 71.008263 samples/sec	 accuracy=66.088068	 loss=1.383325	 lr=0.001000
Epoch[056] Batch [0599]/[3760]	Speed: 69.739025 samples/sec	 accuracy=66.091146	 loss=1.383174	 lr=0.001000
Epoch[056] Batch [0649]/[3760]	Speed: 71.011339 samples/sec	 accuracy=66.048077	 loss=1.386126	 lr=0.001000
Epoch[056] Batch [0699]/[3760]	Speed: 69.612187 samples/sec	 accuracy=65.881696	 loss=1.389066	 lr=0.001000
Epoch[056] Batch [0749]/[3760]	Speed: 69.797871 samples/sec	 accuracy=65.835417	 loss=1.391741	 lr=0.001000
Epoch[056] Batch [0799]/[3760]	Speed: 67.695582 samples/sec	 accuracy=65.878906	 loss=1.390430	 lr=0.001000
Epoch[056] Batch [0849]/[3760]	Speed: 71.101516 samples/sec	 accuracy=65.926471	 loss=1.388770	 lr=0.001000
Epoch[056] Batch [0899]/[3760]	Speed: 69.848340 samples/sec	 accuracy=65.923611	 loss=1.387751	 lr=0.001000
Epoch[056] Batch [0949]/[3760]	Speed: 69.605468 samples/sec	 accuracy=65.901316	 loss=1.387582	 lr=0.001000
Epoch[056] Batch [0999]/[3760]	Speed: 70.711699 samples/sec	 accuracy=65.842188	 loss=1.389638	 lr=0.001000
Epoch[056] Batch [1049]/[3760]	Speed: 67.204954 samples/sec	 accuracy=65.897321	 loss=1.387242	 lr=0.001000
Epoch[056] Batch [1099]/[3760]	Speed: 70.760161 samples/sec	 accuracy=65.894886	 loss=1.388383	 lr=0.001000
Epoch[056] Batch [1149]/[3760]	Speed: 68.374322 samples/sec	 accuracy=65.891304	 loss=1.389853	 lr=0.001000
Epoch[056] Batch [1199]/[3760]	Speed: 70.875908 samples/sec	 accuracy=65.919271	 loss=1.388141	 lr=0.001000
Epoch[056] Batch [1249]/[3760]	Speed: 69.259327 samples/sec	 accuracy=65.976250	 loss=1.385611	 lr=0.001000
Epoch[056] Batch [1299]/[3760]	Speed: 69.075657 samples/sec	 accuracy=66.013221	 loss=1.383487	 lr=0.001000
Epoch[056] Batch [1349]/[3760]	Speed: 71.251908 samples/sec	 accuracy=66.064815	 loss=1.382966	 lr=0.001000
Epoch[056] Batch [1399]/[3760]	Speed: 69.517749 samples/sec	 accuracy=66.071429	 loss=1.382945	 lr=0.001000
Epoch[056] Batch [1449]/[3760]	Speed: 69.197198 samples/sec	 accuracy=66.115302	 loss=1.381530	 lr=0.001000
Epoch[056] Batch [1499]/[3760]	Speed: 68.622004 samples/sec	 accuracy=66.107292	 loss=1.381783	 lr=0.001000
Epoch[056] Batch [1549]/[3760]	Speed: 68.327307 samples/sec	 accuracy=66.097782	 loss=1.382438	 lr=0.001000
Epoch[056] Batch [1599]/[3760]	Speed: 71.472644 samples/sec	 accuracy=66.066406	 loss=1.383343	 lr=0.001000
Epoch[056] Batch [1649]/[3760]	Speed: 69.388160 samples/sec	 accuracy=66.061553	 loss=1.383714	 lr=0.001000
Epoch[056] Batch [1699]/[3760]	Speed: 68.728762 samples/sec	 accuracy=66.056985	 loss=1.385041	 lr=0.001000
Epoch[056] Batch [1749]/[3760]	Speed: 68.376594 samples/sec	 accuracy=66.067857	 loss=1.385497	 lr=0.001000
Epoch[056] Batch [1799]/[3760]	Speed: 72.221832 samples/sec	 accuracy=66.043403	 loss=1.385843	 lr=0.001000
Epoch[056] Batch [1849]/[3760]	Speed: 67.534168 samples/sec	 accuracy=66.016892	 loss=1.387741	 lr=0.001000
Epoch[056] Batch [1899]/[3760]	Speed: 70.465000 samples/sec	 accuracy=66.017270	 loss=1.387843	 lr=0.001000
Epoch[056] Batch [1949]/[3760]	Speed: 68.500645 samples/sec	 accuracy=66.019231	 loss=1.387320	 lr=0.001000
Epoch[056] Batch [1999]/[3760]	Speed: 68.970649 samples/sec	 accuracy=66.045312	 loss=1.386154	 lr=0.001000
Epoch[056] Batch [2049]/[3760]	Speed: 71.237548 samples/sec	 accuracy=66.026677	 loss=1.387268	 lr=0.001000
Epoch[056] Batch [2099]/[3760]	Speed: 69.783613 samples/sec	 accuracy=66.034970	 loss=1.386458	 lr=0.001000
Epoch[056] Batch [2149]/[3760]	Speed: 68.211050 samples/sec	 accuracy=66.038517	 loss=1.385390	 lr=0.001000
Epoch[056] Batch [2199]/[3760]	Speed: 71.136494 samples/sec	 accuracy=66.053977	 loss=1.384948	 lr=0.001000
Epoch[056] Batch [2249]/[3760]	Speed: 68.959138 samples/sec	 accuracy=66.025694	 loss=1.385837	 lr=0.001000
Epoch[056] Batch [2299]/[3760]	Speed: 70.756113 samples/sec	 accuracy=66.005435	 loss=1.386553	 lr=0.001000
Epoch[056] Batch [2349]/[3760]	Speed: 68.585914 samples/sec	 accuracy=66.010638	 loss=1.387119	 lr=0.001000
Epoch[056] Batch [2399]/[3760]	Speed: 70.658572 samples/sec	 accuracy=66.026042	 loss=1.386350	 lr=0.001000
Epoch[056] Batch [2449]/[3760]	Speed: 69.917099 samples/sec	 accuracy=66.018495	 loss=1.386734	 lr=0.001000
Epoch[056] Batch [2499]/[3760]	Speed: 68.037100 samples/sec	 accuracy=66.030000	 loss=1.386676	 lr=0.001000
Epoch[056] Batch [2549]/[3760]	Speed: 71.195578 samples/sec	 accuracy=66.018995	 loss=1.387082	 lr=0.001000
Epoch[056] Batch [2599]/[3760]	Speed: 68.768617 samples/sec	 accuracy=66.041466	 loss=1.386628	 lr=0.001000
Epoch[056] Batch [2649]/[3760]	Speed: 70.407128 samples/sec	 accuracy=66.051297	 loss=1.386474	 lr=0.001000
Epoch[056] Batch [2699]/[3760]	Speed: 70.574950 samples/sec	 accuracy=66.051505	 loss=1.386343	 lr=0.001000
Epoch[056] Batch [2749]/[3760]	Speed: 69.630869 samples/sec	 accuracy=66.051705	 loss=1.386975	 lr=0.001000
Epoch[056] Batch [2799]/[3760]	Speed: 68.942803 samples/sec	 accuracy=66.035714	 loss=1.387291	 lr=0.001000
Epoch[056] Batch [2849]/[3760]	Speed: 69.735938 samples/sec	 accuracy=66.036732	 loss=1.387892	 lr=0.001000
Epoch[056] Batch [2899]/[3760]	Speed: 69.105775 samples/sec	 accuracy=66.045259	 loss=1.387845	 lr=0.001000
Epoch[056] Batch [2949]/[3760]	Speed: 69.567893 samples/sec	 accuracy=66.060381	 loss=1.387356	 lr=0.001000
Epoch[056] Batch [2999]/[3760]	Speed: 70.383972 samples/sec	 accuracy=66.081250	 loss=1.387093	 lr=0.001000
Epoch[056] Batch [3049]/[3760]	Speed: 64.583665 samples/sec	 accuracy=66.107582	 loss=1.386141	 lr=0.001000
Epoch[056] Batch [3099]/[3760]	Speed: 73.969916 samples/sec	 accuracy=66.105847	 loss=1.386356	 lr=0.001000
Epoch[056] Batch [3149]/[3760]	Speed: 69.586802 samples/sec	 accuracy=66.102679	 loss=1.386461	 lr=0.001000
Epoch[056] Batch [3199]/[3760]	Speed: 70.099796 samples/sec	 accuracy=66.071289	 loss=1.387238	 lr=0.001000
Epoch[056] Batch [3249]/[3760]	Speed: 70.270586 samples/sec	 accuracy=66.061538	 loss=1.388166	 lr=0.001000
Epoch[056] Batch [3299]/[3760]	Speed: 71.114457 samples/sec	 accuracy=66.063447	 loss=1.388201	 lr=0.001000
Epoch[056] Batch [3349]/[3760]	Speed: 67.970153 samples/sec	 accuracy=66.073228	 loss=1.388149	 lr=0.001000
Epoch[056] Batch [3399]/[3760]	Speed: 70.590570 samples/sec	 accuracy=66.072151	 loss=1.388901	 lr=0.001000
Epoch[056] Batch [3449]/[3760]	Speed: 67.358859 samples/sec	 accuracy=66.069293	 loss=1.388975	 lr=0.001000
Epoch[056] Batch [3499]/[3760]	Speed: 72.372172 samples/sec	 accuracy=66.074107	 loss=1.388381	 lr=0.001000
Epoch[056] Batch [3549]/[3760]	Speed: 70.132307 samples/sec	 accuracy=66.062060	 loss=1.389012	 lr=0.001000
Epoch[056] Batch [3599]/[3760]	Speed: 69.665152 samples/sec	 accuracy=66.067274	 loss=1.389005	 lr=0.001000
Epoch[056] Batch [3649]/[3760]	Speed: 68.105108 samples/sec	 accuracy=66.073202	 loss=1.388793	 lr=0.001000
Epoch[056] Batch [3699]/[3760]	Speed: 72.017295 samples/sec	 accuracy=66.059544	 loss=1.389340	 lr=0.001000
Epoch[056] Batch [3749]/[3760]	Speed: 78.123844 samples/sec	 accuracy=66.038750	 loss=1.389938	 lr=0.001000
Batch [0049]/[0303]: acc-top1=61.562500 acc-top5=84.031250
Batch [0099]/[0303]: acc-top1=61.703125 acc-top5=83.468750
Batch [0149]/[0303]: acc-top1=62.614583 acc-top5=83.958333
Batch [0199]/[0303]: acc-top1=63.132812 acc-top5=84.375000
Batch [0249]/[0303]: acc-top1=63.212500 acc-top5=84.531250
Batch [0299]/[0303]: acc-top1=63.000000 acc-top5=84.375000
[Epoch 056] training: accuracy=66.043467	 loss=1.389734
[Epoch 056] speed: 69 samples/sec	time cost: 3754.810975
[Epoch 056] validation: acc-top1=62.953795 acc-top5=84.385314 loss=1.602101
Epoch[057] Batch [0049]/[3760]	Speed: 47.645156 samples/sec	 accuracy=65.750000	 loss=1.397732	 lr=0.001000
Epoch[057] Batch [0099]/[3760]	Speed: 69.048984 samples/sec	 accuracy=66.812500	 loss=1.358044	 lr=0.001000
Epoch[057] Batch [0149]/[3760]	Speed: 69.653004 samples/sec	 accuracy=66.583333	 loss=1.361113	 lr=0.001000
Epoch[057] Batch [0199]/[3760]	Speed: 70.223035 samples/sec	 accuracy=66.664062	 loss=1.367214	 lr=0.001000
Epoch[057] Batch [0249]/[3760]	Speed: 68.689743 samples/sec	 accuracy=67.181250	 loss=1.354483	 lr=0.001000
Epoch[057] Batch [0299]/[3760]	Speed: 70.243552 samples/sec	 accuracy=66.927083	 loss=1.357409	 lr=0.001000
Epoch[057] Batch [0349]/[3760]	Speed: 69.071969 samples/sec	 accuracy=66.901786	 loss=1.357328	 lr=0.001000
Epoch[057] Batch [0399]/[3760]	Speed: 70.370315 samples/sec	 accuracy=66.773438	 loss=1.361708	 lr=0.001000
Epoch[057] Batch [0449]/[3760]	Speed: 68.130044 samples/sec	 accuracy=66.854167	 loss=1.359777	 lr=0.001000
Epoch[057] Batch [0499]/[3760]	Speed: 70.353392 samples/sec	 accuracy=66.746875	 loss=1.363482	 lr=0.001000
Epoch[057] Batch [0549]/[3760]	Speed: 69.831562 samples/sec	 accuracy=66.678977	 loss=1.364604	 lr=0.001000
Epoch[057] Batch [0599]/[3760]	Speed: 68.306121 samples/sec	 accuracy=66.539062	 loss=1.369199	 lr=0.001000
Epoch[057] Batch [0649]/[3760]	Speed: 71.297613 samples/sec	 accuracy=66.492788	 loss=1.372998	 lr=0.001000
Epoch[057] Batch [0699]/[3760]	Speed: 69.081905 samples/sec	 accuracy=66.482143	 loss=1.372171	 lr=0.001000
Epoch[057] Batch [0749]/[3760]	Speed: 70.987725 samples/sec	 accuracy=66.437500	 loss=1.373243	 lr=0.001000
Epoch[057] Batch [0799]/[3760]	Speed: 68.134539 samples/sec	 accuracy=66.294922	 loss=1.378631	 lr=0.001000
Epoch[057] Batch [0849]/[3760]	Speed: 70.721718 samples/sec	 accuracy=66.387868	 loss=1.375559	 lr=0.001000
Epoch[057] Batch [0899]/[3760]	Speed: 69.672974 samples/sec	 accuracy=66.362847	 loss=1.378871	 lr=0.001000
Epoch[057] Batch [0949]/[3760]	Speed: 67.913084 samples/sec	 accuracy=66.394737	 loss=1.376400	 lr=0.001000
Epoch[057] Batch [0999]/[3760]	Speed: 70.277937 samples/sec	 accuracy=66.398438	 loss=1.375729	 lr=0.001000
Epoch[057] Batch [1049]/[3760]	Speed: 70.228398 samples/sec	 accuracy=66.364583	 loss=1.379039	 lr=0.001000
Epoch[057] Batch [1099]/[3760]	Speed: 69.715173 samples/sec	 accuracy=66.369318	 loss=1.376551	 lr=0.001000
Epoch[057] Batch [1149]/[3760]	Speed: 69.790016 samples/sec	 accuracy=66.392663	 loss=1.375889	 lr=0.001000
Epoch[057] Batch [1199]/[3760]	Speed: 70.223003 samples/sec	 accuracy=66.432292	 loss=1.374180	 lr=0.001000
Epoch[057] Batch [1249]/[3760]	Speed: 67.900650 samples/sec	 accuracy=66.427500	 loss=1.374967	 lr=0.001000
Epoch[057] Batch [1299]/[3760]	Speed: 69.696354 samples/sec	 accuracy=66.384615	 loss=1.376883	 lr=0.001000
Epoch[057] Batch [1349]/[3760]	Speed: 70.178082 samples/sec	 accuracy=66.358796	 loss=1.377060	 lr=0.001000
Epoch[057] Batch [1399]/[3760]	Speed: 69.404189 samples/sec	 accuracy=66.381696	 loss=1.376944	 lr=0.001000
Epoch[057] Batch [1449]/[3760]	Speed: 68.968350 samples/sec	 accuracy=66.391164	 loss=1.377267	 lr=0.001000
Epoch[057] Batch [1499]/[3760]	Speed: 69.752990 samples/sec	 accuracy=66.373958	 loss=1.378852	 lr=0.001000
Epoch[057] Batch [1549]/[3760]	Speed: 70.087156 samples/sec	 accuracy=66.397177	 loss=1.378353	 lr=0.001000
Epoch[057] Batch [1599]/[3760]	Speed: 70.052918 samples/sec	 accuracy=66.407227	 loss=1.377166	 lr=0.001000
Epoch[057] Batch [1649]/[3760]	Speed: 69.887259 samples/sec	 accuracy=66.384470	 loss=1.378249	 lr=0.001000
Epoch[057] Batch [1699]/[3760]	Speed: 69.193317 samples/sec	 accuracy=66.380515	 loss=1.378822	 lr=0.001000
Epoch[057] Batch [1749]/[3760]	Speed: 70.341512 samples/sec	 accuracy=66.374107	 loss=1.379277	 lr=0.001000
Epoch[057] Batch [1799]/[3760]	Speed: 70.034944 samples/sec	 accuracy=66.340278	 loss=1.379579	 lr=0.001000
Epoch[057] Batch [1849]/[3760]	Speed: 70.119887 samples/sec	 accuracy=66.346284	 loss=1.378912	 lr=0.001000
Epoch[057] Batch [1899]/[3760]	Speed: 68.822293 samples/sec	 accuracy=66.361020	 loss=1.378198	 lr=0.001000
Epoch[057] Batch [1949]/[3760]	Speed: 70.354814 samples/sec	 accuracy=66.356571	 loss=1.376907	 lr=0.001000
Epoch[057] Batch [1999]/[3760]	Speed: 67.991255 samples/sec	 accuracy=66.350781	 loss=1.376585	 lr=0.001000
Epoch[057] Batch [2049]/[3760]	Speed: 68.689371 samples/sec	 accuracy=66.309451	 loss=1.376856	 lr=0.001000
Epoch[057] Batch [2099]/[3760]	Speed: 71.341065 samples/sec	 accuracy=66.276786	 loss=1.377738	 lr=0.001000
Epoch[057] Batch [2149]/[3760]	Speed: 70.232663 samples/sec	 accuracy=66.279797	 loss=1.377031	 lr=0.001000
Epoch[057] Batch [2199]/[3760]	Speed: 69.938104 samples/sec	 accuracy=66.279119	 loss=1.377172	 lr=0.001000
Epoch[057] Batch [2249]/[3760]	Speed: 70.178292 samples/sec	 accuracy=66.270139	 loss=1.377569	 lr=0.001000
Epoch[057] Batch [2299]/[3760]	Speed: 69.285522 samples/sec	 accuracy=66.279891	 loss=1.377939	 lr=0.001000
Epoch[057] Batch [2349]/[3760]	Speed: 69.288751 samples/sec	 accuracy=66.260638	 loss=1.378581	 lr=0.001000
Epoch[057] Batch [2399]/[3760]	Speed: 69.616017 samples/sec	 accuracy=66.263672	 loss=1.378645	 lr=0.001000
Epoch[057] Batch [2449]/[3760]	Speed: 70.776139 samples/sec	 accuracy=66.279974	 loss=1.378047	 lr=0.001000
Epoch[057] Batch [2499]/[3760]	Speed: 70.011037 samples/sec	 accuracy=66.273125	 loss=1.378182	 lr=0.001000
Epoch[057] Batch [2549]/[3760]	Speed: 69.603872 samples/sec	 accuracy=66.261642	 loss=1.378401	 lr=0.001000
Epoch[057] Batch [2599]/[3760]	Speed: 67.375513 samples/sec	 accuracy=66.254207	 loss=1.378758	 lr=0.001000
Epoch[057] Batch [2649]/[3760]	Speed: 70.565787 samples/sec	 accuracy=66.224057	 loss=1.378991	 lr=0.001000
Epoch[057] Batch [2699]/[3760]	Speed: 70.211313 samples/sec	 accuracy=66.229745	 loss=1.378701	 lr=0.001000
Epoch[057] Batch [2749]/[3760]	Speed: 70.642793 samples/sec	 accuracy=66.228409	 loss=1.378564	 lr=0.001000
Epoch[057] Batch [2799]/[3760]	Speed: 69.362197 samples/sec	 accuracy=66.231027	 loss=1.377430	 lr=0.001000
Epoch[057] Batch [2849]/[3760]	Speed: 69.008895 samples/sec	 accuracy=66.248355	 loss=1.376988	 lr=0.001000
Epoch[057] Batch [2899]/[3760]	Speed: 70.555354 samples/sec	 accuracy=66.242457	 loss=1.377441	 lr=0.001000
Epoch[057] Batch [2949]/[3760]	Speed: 69.999683 samples/sec	 accuracy=66.249470	 loss=1.376977	 lr=0.001000
Epoch[057] Batch [2999]/[3760]	Speed: 68.673161 samples/sec	 accuracy=66.263021	 loss=1.375941	 lr=0.001000
Epoch[057] Batch [3049]/[3760]	Speed: 70.514482 samples/sec	 accuracy=66.264344	 loss=1.375716	 lr=0.001000
Epoch[057] Batch [3099]/[3760]	Speed: 70.114101 samples/sec	 accuracy=66.237399	 loss=1.376972	 lr=0.001000
Epoch[057] Batch [3149]/[3760]	Speed: 70.112552 samples/sec	 accuracy=66.236607	 loss=1.376797	 lr=0.001000
Epoch[057] Batch [3199]/[3760]	Speed: 70.349222 samples/sec	 accuracy=66.242676	 loss=1.376559	 lr=0.001000
Epoch[057] Batch [3249]/[3760]	Speed: 68.539595 samples/sec	 accuracy=66.239904	 loss=1.376375	 lr=0.001000
Epoch[057] Batch [3299]/[3760]	Speed: 69.490578 samples/sec	 accuracy=66.240530	 loss=1.376755	 lr=0.001000
Epoch[057] Batch [3349]/[3760]	Speed: 70.601464 samples/sec	 accuracy=66.223414	 loss=1.377246	 lr=0.001000
Epoch[057] Batch [3399]/[3760]	Speed: 68.459682 samples/sec	 accuracy=66.221048	 loss=1.377575	 lr=0.001000
Epoch[057] Batch [3449]/[3760]	Speed: 70.618994 samples/sec	 accuracy=66.210598	 loss=1.377377	 lr=0.001000
Epoch[057] Batch [3499]/[3760]	Speed: 69.143039 samples/sec	 accuracy=66.187500	 loss=1.378274	 lr=0.001000
Epoch[057] Batch [3549]/[3760]	Speed: 71.032278 samples/sec	 accuracy=66.180458	 loss=1.379022	 lr=0.001000
Epoch[057] Batch [3599]/[3760]	Speed: 69.646488 samples/sec	 accuracy=66.176215	 loss=1.379067	 lr=0.001000
Epoch[057] Batch [3649]/[3760]	Speed: 69.598600 samples/sec	 accuracy=66.158818	 loss=1.380005	 lr=0.001000
Epoch[057] Batch [3699]/[3760]	Speed: 69.247478 samples/sec	 accuracy=66.160473	 loss=1.380008	 lr=0.001000
Epoch[057] Batch [3749]/[3760]	Speed: 76.874788 samples/sec	 accuracy=66.148750	 loss=1.380175	 lr=0.001000
Batch [0049]/[0303]: acc-top1=61.718750 acc-top5=83.281250
Batch [0099]/[0303]: acc-top1=61.671875 acc-top5=83.281250
Batch [0149]/[0303]: acc-top1=62.656250 acc-top5=83.750000
Batch [0199]/[0303]: acc-top1=62.984375 acc-top5=84.203125
Batch [0249]/[0303]: acc-top1=63.012500 acc-top5=84.425000
Batch [0299]/[0303]: acc-top1=62.765625 acc-top5=84.276042
[Epoch 057] training: accuracy=66.148188	 loss=1.380168
[Epoch 057] speed: 69 samples/sec	time cost: 3754.944498
[Epoch 057] validation: acc-top1=62.747525 acc-top5=84.271865 loss=1.606785
Epoch[058] Batch [0049]/[3759]	Speed: 48.335850 samples/sec	 accuracy=67.500000	 loss=1.329209	 lr=0.001000
Epoch[058] Batch [0099]/[3759]	Speed: 68.885050 samples/sec	 accuracy=67.343750	 loss=1.347218	 lr=0.001000
Epoch[058] Batch [0149]/[3759]	Speed: 69.560268 samples/sec	 accuracy=67.135417	 loss=1.346929	 lr=0.001000
Epoch[058] Batch [0199]/[3759]	Speed: 70.339601 samples/sec	 accuracy=66.992188	 loss=1.349571	 lr=0.001000
Epoch[058] Batch [0249]/[3759]	Speed: 69.184684 samples/sec	 accuracy=66.862500	 loss=1.363175	 lr=0.001000
Epoch[058] Batch [0299]/[3759]	Speed: 69.871127 samples/sec	 accuracy=66.916667	 loss=1.364441	 lr=0.001000
Epoch[058] Batch [0349]/[3759]	Speed: 70.165196 samples/sec	 accuracy=66.808036	 loss=1.368773	 lr=0.001000
Epoch[058] Batch [0399]/[3759]	Speed: 68.910140 samples/sec	 accuracy=66.824219	 loss=1.369461	 lr=0.001000
Epoch[058] Batch [0449]/[3759]	Speed: 70.370392 samples/sec	 accuracy=66.777778	 loss=1.372073	 lr=0.001000
Epoch[058] Batch [0499]/[3759]	Speed: 67.969513 samples/sec	 accuracy=66.759375	 loss=1.372432	 lr=0.001000
Epoch[058] Batch [0549]/[3759]	Speed: 71.391859 samples/sec	 accuracy=66.826705	 loss=1.370989	 lr=0.001000
Epoch[058] Batch [0599]/[3759]	Speed: 68.861363 samples/sec	 accuracy=66.765625	 loss=1.372947	 lr=0.001000
Epoch[058] Batch [0649]/[3759]	Speed: 68.085971 samples/sec	 accuracy=66.802885	 loss=1.370010	 lr=0.001000
Epoch[058] Batch [0699]/[3759]	Speed: 70.353719 samples/sec	 accuracy=66.750000	 loss=1.372335	 lr=0.001000
Epoch[058] Batch [0749]/[3759]	Speed: 70.456454 samples/sec	 accuracy=66.764583	 loss=1.371797	 lr=0.001000
Epoch[058] Batch [0799]/[3759]	Speed: 69.511018 samples/sec	 accuracy=66.755859	 loss=1.370641	 lr=0.001000
Epoch[058] Batch [0849]/[3759]	Speed: 70.410025 samples/sec	 accuracy=66.772059	 loss=1.367136	 lr=0.001000
Epoch[058] Batch [0899]/[3759]	Speed: 70.089417 samples/sec	 accuracy=66.774306	 loss=1.365331	 lr=0.001000
Epoch[058] Batch [0949]/[3759]	Speed: 68.932203 samples/sec	 accuracy=66.830592	 loss=1.363187	 lr=0.001000
Epoch[058] Batch [0999]/[3759]	Speed: 69.999227 samples/sec	 accuracy=66.798437	 loss=1.364824	 lr=0.001000
Epoch[058] Batch [1049]/[3759]	Speed: 70.637890 samples/sec	 accuracy=66.763393	 loss=1.366478	 lr=0.001000
Epoch[058] Batch [1099]/[3759]	Speed: 69.760815 samples/sec	 accuracy=66.705966	 loss=1.364896	 lr=0.001000
Epoch[058] Batch [1149]/[3759]	Speed: 69.308412 samples/sec	 accuracy=66.618207	 loss=1.368504	 lr=0.001000
Epoch[058] Batch [1199]/[3759]	Speed: 69.808942 samples/sec	 accuracy=66.608073	 loss=1.369189	 lr=0.001000
Epoch[058] Batch [1249]/[3759]	Speed: 69.112784 samples/sec	 accuracy=66.578750	 loss=1.369819	 lr=0.001000
Epoch[058] Batch [1299]/[3759]	Speed: 70.982545 samples/sec	 accuracy=66.556490	 loss=1.371048	 lr=0.001000
Epoch[058] Batch [1349]/[3759]	Speed: 69.358408 samples/sec	 accuracy=66.533565	 loss=1.371820	 lr=0.001000
Epoch[058] Batch [1399]/[3759]	Speed: 68.514188 samples/sec	 accuracy=66.505580	 loss=1.372507	 lr=0.001000
Epoch[058] Batch [1449]/[3759]	Speed: 70.342449 samples/sec	 accuracy=66.485991	 loss=1.373357	 lr=0.001000
Epoch[058] Batch [1499]/[3759]	Speed: 68.528012 samples/sec	 accuracy=66.481250	 loss=1.374128	 lr=0.001000
Epoch[058] Batch [1549]/[3759]	Speed: 70.989134 samples/sec	 accuracy=66.525202	 loss=1.371623	 lr=0.001000
Epoch[058] Batch [1599]/[3759]	Speed: 70.085423 samples/sec	 accuracy=66.533203	 loss=1.371150	 lr=0.001000
Epoch[058] Batch [1649]/[3759]	Speed: 68.329523 samples/sec	 accuracy=66.547348	 loss=1.369956	 lr=0.001000
Epoch[058] Batch [1699]/[3759]	Speed: 70.440864 samples/sec	 accuracy=66.549632	 loss=1.369811	 lr=0.001000
Epoch[058] Batch [1749]/[3759]	Speed: 70.922098 samples/sec	 accuracy=66.527679	 loss=1.371014	 lr=0.001000
Epoch[058] Batch [1799]/[3759]	Speed: 69.846242 samples/sec	 accuracy=66.521701	 loss=1.370689	 lr=0.001000
Epoch[058] Batch [1849]/[3759]	Speed: 69.635282 samples/sec	 accuracy=66.500845	 loss=1.370832	 lr=0.001000
Epoch[058] Batch [1899]/[3759]	Speed: 70.569813 samples/sec	 accuracy=66.529605	 loss=1.369314	 lr=0.001000
Epoch[058] Batch [1949]/[3759]	Speed: 70.041887 samples/sec	 accuracy=66.513622	 loss=1.369213	 lr=0.001000
Epoch[058] Batch [1999]/[3759]	Speed: 69.168668 samples/sec	 accuracy=66.485156	 loss=1.369984	 lr=0.001000
Epoch[058] Batch [2049]/[3759]	Speed: 70.340001 samples/sec	 accuracy=66.483994	 loss=1.370618	 lr=0.001000
Epoch[058] Batch [2099]/[3759]	Speed: 68.621371 samples/sec	 accuracy=66.484375	 loss=1.369913	 lr=0.001000
Epoch[058] Batch [2149]/[3759]	Speed: 70.214725 samples/sec	 accuracy=66.508721	 loss=1.369435	 lr=0.001000
Epoch[058] Batch [2199]/[3759]	Speed: 70.642300 samples/sec	 accuracy=66.458097	 loss=1.371383	 lr=0.001000
Epoch[058] Batch [2249]/[3759]	Speed: 69.165140 samples/sec	 accuracy=66.429861	 loss=1.372011	 lr=0.001000
Epoch[058] Batch [2299]/[3759]	Speed: 69.370512 samples/sec	 accuracy=66.440217	 loss=1.371329	 lr=0.001000
Epoch[058] Batch [2349]/[3759]	Speed: 71.438121 samples/sec	 accuracy=66.428191	 loss=1.371407	 lr=0.001000
Epoch[058] Batch [2399]/[3759]	Speed: 67.913491 samples/sec	 accuracy=66.428385	 loss=1.371511	 lr=0.001000
Epoch[058] Batch [2449]/[3759]	Speed: 69.819897 samples/sec	 accuracy=66.438776	 loss=1.370577	 lr=0.001000
Epoch[058] Batch [2499]/[3759]	Speed: 70.026095 samples/sec	 accuracy=66.445000	 loss=1.370377	 lr=0.001000
Epoch[058] Batch [2549]/[3759]	Speed: 70.301772 samples/sec	 accuracy=66.484069	 loss=1.368997	 lr=0.001000
Epoch[058] Batch [2599]/[3759]	Speed: 69.853787 samples/sec	 accuracy=66.467548	 loss=1.369409	 lr=0.001000
Epoch[058] Batch [2649]/[3759]	Speed: 70.295962 samples/sec	 accuracy=66.451061	 loss=1.369732	 lr=0.001000
Epoch[058] Batch [2699]/[3759]	Speed: 68.398027 samples/sec	 accuracy=66.437500	 loss=1.370331	 lr=0.001000
Epoch[058] Batch [2749]/[3759]	Speed: 70.853949 samples/sec	 accuracy=66.434091	 loss=1.370552	 lr=0.001000
Epoch[058] Batch [2799]/[3759]	Speed: 70.511484 samples/sec	 accuracy=66.452567	 loss=1.370751	 lr=0.001000
Epoch[058] Batch [2849]/[3759]	Speed: 68.349544 samples/sec	 accuracy=66.456140	 loss=1.370782	 lr=0.001000
Epoch[058] Batch [2899]/[3759]	Speed: 70.137150 samples/sec	 accuracy=66.448815	 loss=1.371279	 lr=0.001000
Epoch[058] Batch [2949]/[3759]	Speed: 70.004974 samples/sec	 accuracy=66.430614	 loss=1.372485	 lr=0.001000
Epoch[058] Batch [2999]/[3759]	Speed: 70.766628 samples/sec	 accuracy=66.425000	 loss=1.372806	 lr=0.001000
Epoch[058] Batch [3049]/[3759]	Speed: 67.782295 samples/sec	 accuracy=66.415471	 loss=1.373181	 lr=0.001000
Epoch[058] Batch [3099]/[3759]	Speed: 70.890863 samples/sec	 accuracy=66.412802	 loss=1.373587	 lr=0.001000
Epoch[058] Batch [3149]/[3759]	Speed: 70.894388 samples/sec	 accuracy=66.390873	 loss=1.374099	 lr=0.001000
Epoch[058] Batch [3199]/[3759]	Speed: 69.701705 samples/sec	 accuracy=66.402344	 loss=1.374200	 lr=0.001000
Epoch[058] Batch [3249]/[3759]	Speed: 69.633520 samples/sec	 accuracy=66.386058	 loss=1.374868	 lr=0.001000
Epoch[058] Batch [3299]/[3759]	Speed: 70.163192 samples/sec	 accuracy=66.410985	 loss=1.373847	 lr=0.001000
Epoch[058] Batch [3349]/[3759]	Speed: 69.167048 samples/sec	 accuracy=66.412313	 loss=1.373703	 lr=0.001000
Epoch[058] Batch [3399]/[3759]	Speed: 69.880119 samples/sec	 accuracy=66.410846	 loss=1.373629	 lr=0.001000
Epoch[058] Batch [3449]/[3759]	Speed: 66.690279 samples/sec	 accuracy=66.395380	 loss=1.374560	 lr=0.001000
Epoch[058] Batch [3499]/[3759]	Speed: 72.739599 samples/sec	 accuracy=66.400893	 loss=1.374109	 lr=0.001000
Epoch[058] Batch [3549]/[3759]	Speed: 67.309787 samples/sec	 accuracy=66.408011	 loss=1.373808	 lr=0.001000
Epoch[058] Batch [3599]/[3759]	Speed: 71.039983 samples/sec	 accuracy=66.411458	 loss=1.373867	 lr=0.001000
Epoch[058] Batch [3649]/[3759]	Speed: 69.506253 samples/sec	 accuracy=66.413527	 loss=1.373664	 lr=0.001000
Epoch[058] Batch [3699]/[3759]	Speed: 71.016915 samples/sec	 accuracy=66.413007	 loss=1.373726	 lr=0.001000
Epoch[058] Batch [3749]/[3759]	Speed: 77.235045 samples/sec	 accuracy=66.401250	 loss=1.374151	 lr=0.001000
Batch [0049]/[0303]: acc-top1=62.031250 acc-top5=83.468750
Batch [0099]/[0303]: acc-top1=62.156250 acc-top5=83.437500
Batch [0149]/[0303]: acc-top1=62.958333 acc-top5=83.895833
Batch [0199]/[0303]: acc-top1=63.250000 acc-top5=84.320312
Batch [0249]/[0303]: acc-top1=63.325000 acc-top5=84.525000
Batch [0299]/[0303]: acc-top1=63.052083 acc-top5=84.322917
[Epoch 058] training: accuracy=66.391909	 loss=1.374477
[Epoch 058] speed: 69 samples/sec	time cost: 3751.374909
[Epoch 058] validation: acc-top1=63.015677 acc-top5=84.313119 loss=1.602246
Epoch[059] Batch [0049]/[3760]	Speed: 48.097164 samples/sec	 accuracy=65.906250	 loss=1.390382	 lr=0.001000
Epoch[059] Batch [0099]/[3760]	Speed: 66.320335 samples/sec	 accuracy=66.562500	 loss=1.359493	 lr=0.001000
Epoch[059] Batch [0149]/[3760]	Speed: 69.221905 samples/sec	 accuracy=66.864583	 loss=1.357093	 lr=0.001000
Epoch[059] Batch [0199]/[3760]	Speed: 71.281310 samples/sec	 accuracy=66.664062	 loss=1.368823	 lr=0.001000
Epoch[059] Batch [0249]/[3760]	Speed: 69.526247 samples/sec	 accuracy=66.693750	 loss=1.359069	 lr=0.001000
Epoch[059] Batch [0299]/[3760]	Speed: 66.674768 samples/sec	 accuracy=66.677083	 loss=1.360217	 lr=0.001000
Epoch[059] Batch [0349]/[3760]	Speed: 72.042255 samples/sec	 accuracy=66.683036	 loss=1.360512	 lr=0.001000
Epoch[059] Batch [0399]/[3760]	Speed: 67.940774 samples/sec	 accuracy=66.882812	 loss=1.353115	 lr=0.001000
Epoch[059] Batch [0449]/[3760]	Speed: 71.646522 samples/sec	 accuracy=66.777778	 loss=1.356339	 lr=0.001000
Epoch[059] Batch [0499]/[3760]	Speed: 70.875074 samples/sec	 accuracy=66.737500	 loss=1.356321	 lr=0.001000
Epoch[059] Batch [0549]/[3760]	Speed: 68.563470 samples/sec	 accuracy=66.747159	 loss=1.358563	 lr=0.001000
Epoch[059] Batch [0599]/[3760]	Speed: 70.838808 samples/sec	 accuracy=66.682292	 loss=1.360006	 lr=0.001000
Epoch[059] Batch [0649]/[3760]	Speed: 69.882158 samples/sec	 accuracy=66.649038	 loss=1.362400	 lr=0.001000
Epoch[059] Batch [0699]/[3760]	Speed: 69.977773 samples/sec	 accuracy=66.678571	 loss=1.362892	 lr=0.001000
Epoch[059] Batch [0749]/[3760]	Speed: 68.533912 samples/sec	 accuracy=66.689583	 loss=1.360898	 lr=0.001000
Epoch[059] Batch [0799]/[3760]	Speed: 69.388412 samples/sec	 accuracy=66.703125	 loss=1.359517	 lr=0.001000
Epoch[059] Batch [0849]/[3760]	Speed: 66.012402 samples/sec	 accuracy=66.612132	 loss=1.361400	 lr=0.001000
Epoch[059] Batch [0899]/[3760]	Speed: 71.234282 samples/sec	 accuracy=66.595486	 loss=1.362401	 lr=0.001000
Epoch[059] Batch [0949]/[3760]	Speed: 71.192274 samples/sec	 accuracy=66.580592	 loss=1.362734	 lr=0.001000
Epoch[059] Batch [0999]/[3760]	Speed: 68.821894 samples/sec	 accuracy=66.562500	 loss=1.363956	 lr=0.001000
Epoch[059] Batch [1049]/[3760]	Speed: 70.016233 samples/sec	 accuracy=66.537202	 loss=1.363767	 lr=0.001000
Epoch[059] Batch [1099]/[3760]	Speed: 69.193914 samples/sec	 accuracy=66.565341	 loss=1.362614	 lr=0.001000
Epoch[059] Batch [1149]/[3760]	Speed: 70.188977 samples/sec	 accuracy=66.543478	 loss=1.363046	 lr=0.001000
Epoch[059] Batch [1199]/[3760]	Speed: 69.717072 samples/sec	 accuracy=66.533854	 loss=1.363227	 lr=0.001000
Epoch[059] Batch [1249]/[3760]	Speed: 69.602989 samples/sec	 accuracy=66.531250	 loss=1.363235	 lr=0.001000
Epoch[059] Batch [1299]/[3760]	Speed: 69.907955 samples/sec	 accuracy=66.526442	 loss=1.364719	 lr=0.001000
Epoch[059] Batch [1349]/[3760]	Speed: 69.642414 samples/sec	 accuracy=66.553241	 loss=1.364031	 lr=0.001000
Epoch[059] Batch [1399]/[3760]	Speed: 69.876458 samples/sec	 accuracy=66.550223	 loss=1.364665	 lr=0.001000
Epoch[059] Batch [1449]/[3760]	Speed: 69.167871 samples/sec	 accuracy=66.563578	 loss=1.364796	 lr=0.001000
Epoch[059] Batch [1499]/[3760]	Speed: 69.581540 samples/sec	 accuracy=66.584375	 loss=1.362989	 lr=0.001000
Epoch[059] Batch [1549]/[3760]	Speed: 69.782169 samples/sec	 accuracy=66.596774	 loss=1.363394	 lr=0.001000
Epoch[059] Batch [1599]/[3760]	Speed: 68.389658 samples/sec	 accuracy=66.560547	 loss=1.365518	 lr=0.001000
Epoch[059] Batch [1649]/[3760]	Speed: 70.531060 samples/sec	 accuracy=66.580492	 loss=1.364440	 lr=0.001000
Epoch[059] Batch [1699]/[3760]	Speed: 68.476765 samples/sec	 accuracy=66.582721	 loss=1.364091	 lr=0.001000
Epoch[059] Batch [1749]/[3760]	Speed: 72.088656 samples/sec	 accuracy=66.579464	 loss=1.363687	 lr=0.001000
Epoch[059] Batch [1799]/[3760]	Speed: 69.868001 samples/sec	 accuracy=66.583333	 loss=1.363196	 lr=0.001000
Epoch[059] Batch [1849]/[3760]	Speed: 70.325053 samples/sec	 accuracy=66.597128	 loss=1.362906	 lr=0.001000
Epoch[059] Batch [1899]/[3760]	Speed: 67.650732 samples/sec	 accuracy=66.585526	 loss=1.363574	 lr=0.001000
Epoch[059] Batch [1949]/[3760]	Speed: 69.518868 samples/sec	 accuracy=66.572115	 loss=1.363835	 lr=0.001000
Epoch[059] Batch [1999]/[3760]	Speed: 71.422152 samples/sec	 accuracy=66.598438	 loss=1.363118	 lr=0.001000
Epoch[059] Batch [2049]/[3760]	Speed: 69.226026 samples/sec	 accuracy=66.599085	 loss=1.363288	 lr=0.001000
Epoch[059] Batch [2099]/[3760]	Speed: 69.166511 samples/sec	 accuracy=66.607887	 loss=1.362991	 lr=0.001000
Epoch[059] Batch [2149]/[3760]	Speed: 70.866718 samples/sec	 accuracy=66.611192	 loss=1.362627	 lr=0.001000
Epoch[059] Batch [2199]/[3760]	Speed: 68.849525 samples/sec	 accuracy=66.580256	 loss=1.364071	 lr=0.001000
Epoch[059] Batch [2249]/[3760]	Speed: 69.497105 samples/sec	 accuracy=66.560417	 loss=1.364490	 lr=0.001000
Epoch[059] Batch [2299]/[3760]	Speed: 70.750682 samples/sec	 accuracy=66.536005	 loss=1.364986	 lr=0.001000
Epoch[059] Batch [2349]/[3760]	Speed: 69.092322 samples/sec	 accuracy=66.519282	 loss=1.365089	 lr=0.001000
Epoch[059] Batch [2399]/[3760]	Speed: 69.598862 samples/sec	 accuracy=66.536458	 loss=1.365096	 lr=0.001000
Epoch[059] Batch [2449]/[3760]	Speed: 70.427063 samples/sec	 accuracy=66.550383	 loss=1.364683	 lr=0.001000
Epoch[059] Batch [2499]/[3760]	Speed: 66.307303 samples/sec	 accuracy=66.549375	 loss=1.364725	 lr=0.001000
Epoch[059] Batch [2549]/[3760]	Speed: 73.370814 samples/sec	 accuracy=66.570466	 loss=1.364196	 lr=0.001000
Epoch[059] Batch [2599]/[3760]	Speed: 68.195183 samples/sec	 accuracy=66.578726	 loss=1.363589	 lr=0.001000
Epoch[059] Batch [2649]/[3760]	Speed: 70.614489 samples/sec	 accuracy=66.576651	 loss=1.363838	 lr=0.001000
Epoch[059] Batch [2699]/[3760]	Speed: 68.882027 samples/sec	 accuracy=66.583912	 loss=1.363092	 lr=0.001000
Epoch[059] Batch [2749]/[3760]	Speed: 69.904280 samples/sec	 accuracy=66.572727	 loss=1.363957	 lr=0.001000
Epoch[059] Batch [2799]/[3760]	Speed: 69.886557 samples/sec	 accuracy=66.572545	 loss=1.363504	 lr=0.001000
Epoch[059] Batch [2849]/[3760]	Speed: 69.673119 samples/sec	 accuracy=66.596491	 loss=1.362659	 lr=0.001000
Epoch[059] Batch [2899]/[3760]	Speed: 69.302319 samples/sec	 accuracy=66.589978	 loss=1.363316	 lr=0.001000
Epoch[059] Batch [2949]/[3760]	Speed: 69.182306 samples/sec	 accuracy=66.590042	 loss=1.363353	 lr=0.001000
Epoch[059] Batch [2999]/[3760]	Speed: 69.187087 samples/sec	 accuracy=66.593229	 loss=1.363039	 lr=0.001000
Epoch[059] Batch [3049]/[3760]	Speed: 68.369004 samples/sec	 accuracy=66.571209	 loss=1.363531	 lr=0.001000
Epoch[059] Batch [3099]/[3760]	Speed: 70.642067 samples/sec	 accuracy=66.587702	 loss=1.363584	 lr=0.001000
Epoch[059] Batch [3149]/[3760]	Speed: 68.754533 samples/sec	 accuracy=66.542659	 loss=1.364756	 lr=0.001000
Epoch[059] Batch [3199]/[3760]	Speed: 68.988830 samples/sec	 accuracy=66.529297	 loss=1.365432	 lr=0.001000
Epoch[059] Batch [3249]/[3760]	Speed: 66.572996 samples/sec	 accuracy=66.528846	 loss=1.365265	 lr=0.001000
Epoch[059] Batch [3299]/[3760]	Speed: 74.228742 samples/sec	 accuracy=66.515152	 loss=1.365816	 lr=0.001000
Epoch[059] Batch [3349]/[3760]	Speed: 69.222948 samples/sec	 accuracy=66.513993	 loss=1.365326	 lr=0.001000
Epoch[059] Batch [3399]/[3760]	Speed: 69.767413 samples/sec	 accuracy=66.485294	 loss=1.366791	 lr=0.001000
Epoch[059] Batch [3449]/[3760]	Speed: 69.559589 samples/sec	 accuracy=66.466486	 loss=1.366990	 lr=0.001000
Epoch[059] Batch [3499]/[3760]	Speed: 70.227386 samples/sec	 accuracy=66.454018	 loss=1.367593	 lr=0.001000
Epoch[059] Batch [3549]/[3760]	Speed: 70.790203 samples/sec	 accuracy=66.433979	 loss=1.368250	 lr=0.001000
Epoch[059] Batch [3599]/[3760]	Speed: 68.781537 samples/sec	 accuracy=66.434462	 loss=1.368416	 lr=0.001000
Epoch[059] Batch [3649]/[3760]	Speed: 70.100105 samples/sec	 accuracy=66.444777	 loss=1.368042	 lr=0.001000
Epoch[059] Batch [3699]/[3760]	Speed: 68.399940 samples/sec	 accuracy=66.448057	 loss=1.368055	 lr=0.001000
Epoch[059] Batch [3749]/[3760]	Speed: 77.955875 samples/sec	 accuracy=66.438750	 loss=1.368745	 lr=0.001000
Batch [0049]/[0303]: acc-top1=62.062500 acc-top5=83.656250
Batch [0099]/[0303]: acc-top1=62.031250 acc-top5=83.609375
Batch [0149]/[0303]: acc-top1=62.875000 acc-top5=83.864583
Batch [0199]/[0303]: acc-top1=63.273437 acc-top5=84.359375
Batch [0249]/[0303]: acc-top1=63.212500 acc-top5=84.512500
Batch [0299]/[0303]: acc-top1=63.046875 acc-top5=84.348958
[Epoch 059] training: accuracy=66.438248	 loss=1.368746
[Epoch 059] speed: 69 samples/sec	time cost: 3758.503896
[Epoch 059] validation: acc-top1=63.031147 acc-top5=84.359530 loss=1.602395
Epoch[060] Batch [0049]/[3760]	Speed: 48.178166 samples/sec	 accuracy=67.000000	 loss=1.313878	 lr=0.001000
Epoch[060] Batch [0099]/[3760]	Speed: 68.550802 samples/sec	 accuracy=66.968750	 loss=1.337754	 lr=0.001000
Epoch[060] Batch [0149]/[3760]	Speed: 69.217647 samples/sec	 accuracy=67.125000	 loss=1.345998	 lr=0.001000
Epoch[060] Batch [0199]/[3760]	Speed: 69.195913 samples/sec	 accuracy=66.875000	 loss=1.348918	 lr=0.001000
Epoch[060] Batch [0249]/[3760]	Speed: 69.893132 samples/sec	 accuracy=66.968750	 loss=1.346772	 lr=0.001000
Epoch[060] Batch [0299]/[3760]	Speed: 69.504946 samples/sec	 accuracy=66.619792	 loss=1.358602	 lr=0.001000
Epoch[060] Batch [0349]/[3760]	Speed: 68.265384 samples/sec	 accuracy=66.691964	 loss=1.357348	 lr=0.001000
Epoch[060] Batch [0399]/[3760]	Speed: 69.420291 samples/sec	 accuracy=66.972656	 loss=1.347752	 lr=0.001000
Epoch[060] Batch [0449]/[3760]	Speed: 70.192671 samples/sec	 accuracy=66.934028	 loss=1.346531	 lr=0.001000
Epoch[060] Batch [0499]/[3760]	Speed: 69.875841 samples/sec	 accuracy=66.893750	 loss=1.349050	 lr=0.001000
Epoch[060] Batch [0549]/[3760]	Speed: 69.832473 samples/sec	 accuracy=66.883523	 loss=1.350662	 lr=0.001000
Epoch[060] Batch [0599]/[3760]	Speed: 70.015754 samples/sec	 accuracy=66.869792	 loss=1.352800	 lr=0.001000
Epoch[060] Batch [0649]/[3760]	Speed: 69.297428 samples/sec	 accuracy=66.846154	 loss=1.351796	 lr=0.001000
Epoch[060] Batch [0699]/[3760]	Speed: 70.279877 samples/sec	 accuracy=66.738839	 loss=1.354835	 lr=0.001000
Epoch[060] Batch [0749]/[3760]	Speed: 70.019596 samples/sec	 accuracy=66.683333	 loss=1.356161	 lr=0.001000
Epoch[060] Batch [0799]/[3760]	Speed: 68.618526 samples/sec	 accuracy=66.708984	 loss=1.354644	 lr=0.001000
Epoch[060] Batch [0849]/[3760]	Speed: 69.409800 samples/sec	 accuracy=66.718750	 loss=1.353474	 lr=0.001000
Epoch[060] Batch [0899]/[3760]	Speed: 69.859304 samples/sec	 accuracy=66.694444	 loss=1.354899	 lr=0.001000
Epoch[060] Batch [0949]/[3760]	Speed: 69.936311 samples/sec	 accuracy=66.570724	 loss=1.359359	 lr=0.001000
Epoch[060] Batch [0999]/[3760]	Speed: 68.754302 samples/sec	 accuracy=66.532813	 loss=1.360497	 lr=0.001000
Epoch[060] Batch [1049]/[3760]	Speed: 71.343052 samples/sec	 accuracy=66.544643	 loss=1.359978	 lr=0.001000
Epoch[060] Batch [1099]/[3760]	Speed: 69.646972 samples/sec	 accuracy=66.465909	 loss=1.362268	 lr=0.001000
Epoch[060] Batch [1149]/[3760]	Speed: 68.464260 samples/sec	 accuracy=66.513587	 loss=1.362525	 lr=0.001000
Epoch[060] Batch [1199]/[3760]	Speed: 69.805212 samples/sec	 accuracy=66.476562	 loss=1.364184	 lr=0.001000
Epoch[060] Batch [1249]/[3760]	Speed: 71.222680 samples/sec	 accuracy=66.448750	 loss=1.365719	 lr=0.001000
Epoch[060] Batch [1299]/[3760]	Speed: 70.286256 samples/sec	 accuracy=66.513221	 loss=1.363819	 lr=0.001000
Epoch[060] Batch [1349]/[3760]	Speed: 68.520685 samples/sec	 accuracy=66.516204	 loss=1.363605	 lr=0.001000
Epoch[060] Batch [1399]/[3760]	Speed: 70.307628 samples/sec	 accuracy=66.599330	 loss=1.360378	 lr=0.001000
Epoch[060] Batch [1449]/[3760]	Speed: 69.025930 samples/sec	 accuracy=66.578664	 loss=1.361558	 lr=0.001000
Epoch[060] Batch [1499]/[3760]	Speed: 69.862825 samples/sec	 accuracy=66.551042	 loss=1.362282	 lr=0.001000
Epoch[060] Batch [1549]/[3760]	Speed: 69.679850 samples/sec	 accuracy=66.525202	 loss=1.362693	 lr=0.001000
Epoch[060] Batch [1599]/[3760]	Speed: 69.577856 samples/sec	 accuracy=66.530273	 loss=1.362521	 lr=0.001000
Epoch[060] Batch [1649]/[3760]	Speed: 70.128212 samples/sec	 accuracy=66.519886	 loss=1.363024	 lr=0.001000
Epoch[060] Batch [1699]/[3760]	Speed: 69.415371 samples/sec	 accuracy=66.544118	 loss=1.362652	 lr=0.001000
Epoch[060] Batch [1749]/[3760]	Speed: 69.988223 samples/sec	 accuracy=66.562500	 loss=1.362030	 lr=0.001000
Epoch[060] Batch [1799]/[3760]	Speed: 69.530985 samples/sec	 accuracy=66.609375	 loss=1.360568	 lr=0.001000
Epoch[060] Batch [1849]/[3760]	Speed: 68.622265 samples/sec	 accuracy=66.627534	 loss=1.359680	 lr=0.001000
Epoch[060] Batch [1899]/[3760]	Speed: 70.502038 samples/sec	 accuracy=66.623355	 loss=1.360170	 lr=0.001000
Epoch[060] Batch [1949]/[3760]	Speed: 70.268375 samples/sec	 accuracy=66.637821	 loss=1.359471	 lr=0.001000
Epoch[060] Batch [1999]/[3760]	Speed: 70.106609 samples/sec	 accuracy=66.644531	 loss=1.360769	 lr=0.001000
Epoch[060] Batch [2049]/[3760]	Speed: 69.585153 samples/sec	 accuracy=66.632622	 loss=1.361949	 lr=0.001000
Epoch[060] Batch [2099]/[3760]	Speed: 68.768121 samples/sec	 accuracy=66.581101	 loss=1.364076	 lr=0.001000
Epoch[060] Batch [2149]/[3760]	Speed: 70.746278 samples/sec	 accuracy=66.581395	 loss=1.363962	 lr=0.001000
Epoch[060] Batch [2199]/[3760]	Speed: 69.247315 samples/sec	 accuracy=66.586648	 loss=1.364343	 lr=0.001000
Epoch[060] Batch [2249]/[3760]	Speed: 71.142743 samples/sec	 accuracy=66.608333	 loss=1.363048	 lr=0.001000
Epoch[060] Batch [2299]/[3760]	Speed: 65.850276 samples/sec	 accuracy=66.602582	 loss=1.362921	 lr=0.001000
Epoch[060] Batch [2349]/[3760]	Speed: 72.821544 samples/sec	 accuracy=66.626995	 loss=1.362750	 lr=0.001000
Epoch[060] Batch [2399]/[3760]	Speed: 69.514232 samples/sec	 accuracy=66.606120	 loss=1.363241	 lr=0.001000
Epoch[060] Batch [2449]/[3760]	Speed: 69.545603 samples/sec	 accuracy=66.633929	 loss=1.362330	 lr=0.001000
Epoch[060] Batch [2499]/[3760]	Speed: 70.290496 samples/sec	 accuracy=66.642500	 loss=1.362464	 lr=0.001000
Epoch[060] Batch [2549]/[3760]	Speed: 69.583663 samples/sec	 accuracy=66.620098	 loss=1.362820	 lr=0.001000
Epoch[060] Batch [2599]/[3760]	Speed: 69.627857 samples/sec	 accuracy=66.624399	 loss=1.362632	 lr=0.001000
Epoch[060] Batch [2649]/[3760]	Speed: 70.511455 samples/sec	 accuracy=66.604953	 loss=1.362523	 lr=0.001000
Epoch[060] Batch [2699]/[3760]	Speed: 68.164435 samples/sec	 accuracy=66.583912	 loss=1.363966	 lr=0.001000
Epoch[060] Batch [2749]/[3760]	Speed: 70.408646 samples/sec	 accuracy=66.579545	 loss=1.363910	 lr=0.001000
Epoch[060] Batch [2799]/[3760]	Speed: 70.753178 samples/sec	 accuracy=66.559152	 loss=1.364542	 lr=0.001000
Epoch[060] Batch [2849]/[3760]	Speed: 69.865897 samples/sec	 accuracy=66.574561	 loss=1.363973	 lr=0.001000
Epoch[060] Batch [2899]/[3760]	Speed: 68.559384 samples/sec	 accuracy=66.577586	 loss=1.363914	 lr=0.001000
Epoch[060] Batch [2949]/[3760]	Speed: 70.175333 samples/sec	 accuracy=66.587394	 loss=1.363683	 lr=0.001000
Epoch[060] Batch [2999]/[3760]	Speed: 69.620940 samples/sec	 accuracy=66.564583	 loss=1.364274	 lr=0.001000
Epoch[060] Batch [3049]/[3760]	Speed: 70.792346 samples/sec	 accuracy=66.552766	 loss=1.364889	 lr=0.001000
Epoch[060] Batch [3099]/[3760]	Speed: 68.681944 samples/sec	 accuracy=66.561996	 loss=1.364911	 lr=0.001000
Epoch[060] Batch [3149]/[3760]	Speed: 69.495373 samples/sec	 accuracy=66.548115	 loss=1.364718	 lr=0.001000
Epoch[060] Batch [3199]/[3760]	Speed: 69.613620 samples/sec	 accuracy=66.556641	 loss=1.364223	 lr=0.001000
Epoch[060] Batch [3249]/[3760]	Speed: 69.986424 samples/sec	 accuracy=66.555288	 loss=1.364099	 lr=0.001000
Epoch[060] Batch [3299]/[3760]	Speed: 69.577778 samples/sec	 accuracy=66.542140	 loss=1.364623	 lr=0.001000
Epoch[060] Batch [3349]/[3760]	Speed: 69.146793 samples/sec	 accuracy=66.522854	 loss=1.365658	 lr=0.001000
Epoch[060] Batch [3399]/[3760]	Speed: 69.679795 samples/sec	 accuracy=66.523897	 loss=1.365726	 lr=0.001000
Epoch[060] Batch [3449]/[3760]	Speed: 69.883183 samples/sec	 accuracy=66.543478	 loss=1.365203	 lr=0.001000
Epoch[060] Batch [3499]/[3760]	Speed: 70.555983 samples/sec	 accuracy=66.562946	 loss=1.364382	 lr=0.001000
Epoch[060] Batch [3549]/[3760]	Speed: 69.692703 samples/sec	 accuracy=66.567342	 loss=1.363908	 lr=0.001000
Epoch[060] Batch [3599]/[3760]	Speed: 69.404721 samples/sec	 accuracy=66.574653	 loss=1.363992	 lr=0.001000
Epoch[060] Batch [3649]/[3760]	Speed: 70.210570 samples/sec	 accuracy=66.576627	 loss=1.363307	 lr=0.001000
Epoch[060] Batch [3699]/[3760]	Speed: 68.526355 samples/sec	 accuracy=66.584882	 loss=1.362692	 lr=0.001000
Epoch[060] Batch [3749]/[3760]	Speed: 77.700647 samples/sec	 accuracy=66.584583	 loss=1.362878	 lr=0.001000
Batch [0049]/[0303]: acc-top1=61.437500 acc-top5=83.812500
Batch [0099]/[0303]: acc-top1=61.765625 acc-top5=83.625000
Batch [0149]/[0303]: acc-top1=62.750000 acc-top5=83.885417
Batch [0199]/[0303]: acc-top1=63.304688 acc-top5=84.304688
Batch [0249]/[0303]: acc-top1=63.362500 acc-top5=84.500000
Batch [0299]/[0303]: acc-top1=63.083333 acc-top5=84.328125
[Epoch 060] training: accuracy=66.596576	 loss=1.362457
[Epoch 060] speed: 69 samples/sec	time cost: 3755.469269
[Epoch 060] validation: acc-top1=63.077558 acc-top5=84.338903 loss=1.599669
Epoch[061] Batch [0049]/[3759]	Speed: 47.457948 samples/sec	 accuracy=66.781250	 loss=1.313421	 lr=0.001000
Epoch[061] Batch [0099]/[3759]	Speed: 68.975172 samples/sec	 accuracy=66.328125	 loss=1.364428	 lr=0.001000
Epoch[061] Batch [0149]/[3759]	Speed: 68.806278 samples/sec	 accuracy=66.552083	 loss=1.357556	 lr=0.001000
Epoch[061] Batch [0199]/[3759]	Speed: 70.056349 samples/sec	 accuracy=66.367188	 loss=1.358786	 lr=0.001000
Epoch[061] Batch [0249]/[3759]	Speed: 70.612787 samples/sec	 accuracy=66.275000	 loss=1.370842	 lr=0.001000
Epoch[061] Batch [0299]/[3759]	Speed: 70.226290 samples/sec	 accuracy=66.437500	 loss=1.361877	 lr=0.001000
Epoch[061] Batch [0349]/[3759]	Speed: 69.589037 samples/sec	 accuracy=66.383929	 loss=1.364100	 lr=0.001000
Epoch[061] Batch [0399]/[3759]	Speed: 69.913554 samples/sec	 accuracy=66.585938	 loss=1.357549	 lr=0.001000
Epoch[061] Batch [0449]/[3759]	Speed: 70.645774 samples/sec	 accuracy=66.607639	 loss=1.358165	 lr=0.001000
Epoch[061] Batch [0499]/[3759]	Speed: 68.165559 samples/sec	 accuracy=66.496875	 loss=1.361422	 lr=0.001000
Epoch[061] Batch [0549]/[3759]	Speed: 71.300875 samples/sec	 accuracy=66.477273	 loss=1.363522	 lr=0.001000
Epoch[061] Batch [0599]/[3759]	Speed: 69.835841 samples/sec	 accuracy=66.466146	 loss=1.362396	 lr=0.001000
Epoch[061] Batch [0649]/[3759]	Speed: 70.286123 samples/sec	 accuracy=66.500000	 loss=1.362854	 lr=0.001000
Epoch[061] Batch [0699]/[3759]	Speed: 69.503520 samples/sec	 accuracy=66.520089	 loss=1.361506	 lr=0.001000
Epoch[061] Batch [0749]/[3759]	Speed: 70.164150 samples/sec	 accuracy=66.539583	 loss=1.361120	 lr=0.001000
Epoch[061] Batch [0799]/[3759]	Speed: 69.170626 samples/sec	 accuracy=66.425781	 loss=1.366668	 lr=0.001000
Epoch[061] Batch [0849]/[3759]	Speed: 70.041363 samples/sec	 accuracy=66.461397	 loss=1.365878	 lr=0.001000
Epoch[061] Batch [0899]/[3759]	Speed: 71.150092 samples/sec	 accuracy=66.437500	 loss=1.365577	 lr=0.001000
Epoch[061] Batch [0949]/[3759]	Speed: 68.752147 samples/sec	 accuracy=66.442434	 loss=1.366706	 lr=0.001000
Epoch[061] Batch [0999]/[3759]	Speed: 69.229422 samples/sec	 accuracy=66.453125	 loss=1.366942	 lr=0.001000
Epoch[061] Batch [1049]/[3759]	Speed: 69.463744 samples/sec	 accuracy=66.494048	 loss=1.364798	 lr=0.001000
Epoch[061] Batch [1099]/[3759]	Speed: 70.494965 samples/sec	 accuracy=66.522727	 loss=1.364815	 lr=0.001000
Epoch[061] Batch [1149]/[3759]	Speed: 69.270846 samples/sec	 accuracy=66.527174	 loss=1.364328	 lr=0.001000
Epoch[061] Batch [1199]/[3759]	Speed: 69.704336 samples/sec	 accuracy=66.536458	 loss=1.364648	 lr=0.001000
Epoch[061] Batch [1249]/[3759]	Speed: 69.840179 samples/sec	 accuracy=66.511250	 loss=1.364847	 lr=0.001000
Epoch[061] Batch [1299]/[3759]	Speed: 69.916082 samples/sec	 accuracy=66.548077	 loss=1.363592	 lr=0.001000
Epoch[061] Batch [1349]/[3759]	Speed: 70.213505 samples/sec	 accuracy=66.581019	 loss=1.361794	 lr=0.001000
Epoch[061] Batch [1399]/[3759]	Speed: 69.359555 samples/sec	 accuracy=66.625000	 loss=1.360067	 lr=0.001000
Epoch[061] Batch [1449]/[3759]	Speed: 70.325379 samples/sec	 accuracy=66.642241	 loss=1.358734	 lr=0.001000
Epoch[061] Batch [1499]/[3759]	Speed: 69.118665 samples/sec	 accuracy=66.586458	 loss=1.360276	 lr=0.001000
Epoch[061] Batch [1549]/[3759]	Speed: 70.720115 samples/sec	 accuracy=66.569556	 loss=1.360920	 lr=0.001000
Epoch[061] Batch [1599]/[3759]	Speed: 68.742400 samples/sec	 accuracy=66.585938	 loss=1.360367	 lr=0.001000
Epoch[061] Batch [1649]/[3759]	Speed: 70.767754 samples/sec	 accuracy=66.621212	 loss=1.360038	 lr=0.001000
Epoch[061] Batch [1699]/[3759]	Speed: 69.845088 samples/sec	 accuracy=66.602941	 loss=1.359472	 lr=0.001000
Epoch[061] Batch [1749]/[3759]	Speed: 67.352858 samples/sec	 accuracy=66.621429	 loss=1.358400	 lr=0.001000
Epoch[061] Batch [1799]/[3759]	Speed: 71.361557 samples/sec	 accuracy=66.677083	 loss=1.356710	 lr=0.001000
Epoch[061] Batch [1849]/[3759]	Speed: 67.848989 samples/sec	 accuracy=66.673986	 loss=1.356644	 lr=0.001000
Epoch[061] Batch [1899]/[3759]	Speed: 71.893650 samples/sec	 accuracy=66.693257	 loss=1.355779	 lr=0.001000
Epoch[061] Batch [1949]/[3759]	Speed: 70.701613 samples/sec	 accuracy=66.693109	 loss=1.355667	 lr=0.001000
Epoch[061] Batch [1999]/[3759]	Speed: 68.016802 samples/sec	 accuracy=66.722656	 loss=1.354674	 lr=0.001000
Epoch[061] Batch [2049]/[3759]	Speed: 70.067520 samples/sec	 accuracy=66.706555	 loss=1.355569	 lr=0.001000
Epoch[061] Batch [2099]/[3759]	Speed: 69.413321 samples/sec	 accuracy=66.722470	 loss=1.354567	 lr=0.001000
Epoch[061] Batch [2149]/[3759]	Speed: 69.802351 samples/sec	 accuracy=66.734738	 loss=1.354636	 lr=0.001000
Epoch[061] Batch [2199]/[3759]	Speed: 68.945334 samples/sec	 accuracy=66.713068	 loss=1.355328	 lr=0.001000
Epoch[061] Batch [2249]/[3759]	Speed: 71.057222 samples/sec	 accuracy=66.710417	 loss=1.355055	 lr=0.001000
Epoch[061] Batch [2299]/[3759]	Speed: 69.694006 samples/sec	 accuracy=66.693614	 loss=1.355615	 lr=0.001000
Epoch[061] Batch [2349]/[3759]	Speed: 70.064871 samples/sec	 accuracy=66.680851	 loss=1.356832	 lr=0.001000
Epoch[061] Batch [2399]/[3759]	Speed: 69.923538 samples/sec	 accuracy=66.680990	 loss=1.356664	 lr=0.001000
Epoch[061] Batch [2449]/[3759]	Speed: 70.976022 samples/sec	 accuracy=66.667730	 loss=1.358030	 lr=0.001000
Epoch[061] Batch [2499]/[3759]	Speed: 68.532555 samples/sec	 accuracy=66.650625	 loss=1.358967	 lr=0.001000
Epoch[061] Batch [2549]/[3759]	Speed: 69.921511 samples/sec	 accuracy=66.639093	 loss=1.359680	 lr=0.001000
Epoch[061] Batch [2599]/[3759]	Speed: 70.685711 samples/sec	 accuracy=66.606370	 loss=1.360630	 lr=0.001000
Epoch[061] Batch [2649]/[3759]	Speed: 68.580207 samples/sec	 accuracy=66.604363	 loss=1.360437	 lr=0.001000
Epoch[061] Batch [2699]/[3759]	Speed: 70.496388 samples/sec	 accuracy=66.604745	 loss=1.360340	 lr=0.001000
Epoch[061] Batch [2749]/[3759]	Speed: 69.272198 samples/sec	 accuracy=66.602841	 loss=1.360610	 lr=0.001000
Epoch[061] Batch [2799]/[3759]	Speed: 71.021344 samples/sec	 accuracy=66.604911	 loss=1.360437	 lr=0.001000
Epoch[061] Batch [2849]/[3759]	Speed: 67.962738 samples/sec	 accuracy=66.614035	 loss=1.360018	 lr=0.001000
Epoch[061] Batch [2899]/[3759]	Speed: 70.605986 samples/sec	 accuracy=66.628233	 loss=1.359731	 lr=0.001000
Epoch[061] Batch [2949]/[3759]	Speed: 68.734302 samples/sec	 accuracy=66.614936	 loss=1.359870	 lr=0.001000
Epoch[061] Batch [2999]/[3759]	Speed: 70.761950 samples/sec	 accuracy=66.606771	 loss=1.360164	 lr=0.001000
Epoch[061] Batch [3049]/[3759]	Speed: 67.288500 samples/sec	 accuracy=66.604508	 loss=1.360924	 lr=0.001000
Epoch[061] Batch [3099]/[3759]	Speed: 70.878519 samples/sec	 accuracy=66.613407	 loss=1.360391	 lr=0.001000
Epoch[061] Batch [3149]/[3759]	Speed: 70.653768 samples/sec	 accuracy=66.615575	 loss=1.360453	 lr=0.001000
Epoch[061] Batch [3199]/[3759]	Speed: 69.274725 samples/sec	 accuracy=66.624512	 loss=1.360215	 lr=0.001000
Epoch[061] Batch [3249]/[3759]	Speed: 71.515430 samples/sec	 accuracy=66.641346	 loss=1.359456	 lr=0.001000
Epoch[061] Batch [3299]/[3759]	Speed: 68.795937 samples/sec	 accuracy=66.633996	 loss=1.359761	 lr=0.001000
Epoch[061] Batch [3349]/[3759]	Speed: 67.828946 samples/sec	 accuracy=66.617071	 loss=1.360769	 lr=0.001000
Epoch[061] Batch [3399]/[3759]	Speed: 70.019136 samples/sec	 accuracy=66.618107	 loss=1.360317	 lr=0.001000
Epoch[061] Batch [3449]/[3759]	Speed: 69.722411 samples/sec	 accuracy=66.605525	 loss=1.360391	 lr=0.001000
Epoch[061] Batch [3499]/[3759]	Speed: 71.245741 samples/sec	 accuracy=66.607589	 loss=1.360644	 lr=0.001000
Epoch[061] Batch [3549]/[3759]	Speed: 69.378113 samples/sec	 accuracy=66.608275	 loss=1.360505	 lr=0.001000
Epoch[061] Batch [3599]/[3759]	Speed: 69.943467 samples/sec	 accuracy=66.611111	 loss=1.360825	 lr=0.001000
Epoch[061] Batch [3649]/[3759]	Speed: 68.600652 samples/sec	 accuracy=66.598887	 loss=1.361584	 lr=0.001000
Epoch[061] Batch [3699]/[3759]	Speed: 70.822809 samples/sec	 accuracy=66.597551	 loss=1.361583	 lr=0.001000
Epoch[061] Batch [3749]/[3759]	Speed: 78.209243 samples/sec	 accuracy=66.593333	 loss=1.361633	 lr=0.001000
Batch [0049]/[0303]: acc-top1=61.906250 acc-top5=83.718750
Batch [0099]/[0303]: acc-top1=62.484375 acc-top5=83.671875
Batch [0149]/[0303]: acc-top1=62.916667 acc-top5=84.239583
Batch [0199]/[0303]: acc-top1=63.164062 acc-top5=84.562500
Batch [0249]/[0303]: acc-top1=63.275000 acc-top5=84.768750
Batch [0299]/[0303]: acc-top1=63.088542 acc-top5=84.546875
[Epoch 061] training: accuracy=66.600991	 loss=1.361496
[Epoch 061] speed: 69 samples/sec	time cost: 3748.993271
[Epoch 061] validation: acc-top1=63.062087 acc-top5=84.555487 loss=1.595635
Epoch[062] Batch [0049]/[3760]	Speed: 47.679981 samples/sec	 accuracy=66.468750	 loss=1.382851	 lr=0.001000
Epoch[062] Batch [0099]/[3760]	Speed: 66.688423 samples/sec	 accuracy=67.609375	 loss=1.337933	 lr=0.001000
Epoch[062] Batch [0149]/[3760]	Speed: 71.139064 samples/sec	 accuracy=67.541667	 loss=1.349113	 lr=0.001000
Epoch[062] Batch [0199]/[3760]	Speed: 69.117483 samples/sec	 accuracy=67.609375	 loss=1.336084	 lr=0.001000
Epoch[062] Batch [0249]/[3760]	Speed: 70.538461 samples/sec	 accuracy=67.537500	 loss=1.338349	 lr=0.001000
Epoch[062] Batch [0299]/[3760]	Speed: 69.305492 samples/sec	 accuracy=67.333333	 loss=1.342767	 lr=0.001000
Epoch[062] Batch [0349]/[3760]	Speed: 70.816046 samples/sec	 accuracy=67.156250	 loss=1.348554	 lr=0.001000
Epoch[062] Batch [0399]/[3760]	Speed: 68.339165 samples/sec	 accuracy=67.085938	 loss=1.349954	 lr=0.001000
Epoch[062] Batch [0449]/[3760]	Speed: 70.713742 samples/sec	 accuracy=67.024306	 loss=1.354375	 lr=0.001000
Epoch[062] Batch [0499]/[3760]	Speed: 70.669589 samples/sec	 accuracy=67.037500	 loss=1.354200	 lr=0.001000
Epoch[062] Batch [0549]/[3760]	Speed: 69.608552 samples/sec	 accuracy=67.014205	 loss=1.353745	 lr=0.001000
Epoch[062] Batch [0599]/[3760]	Speed: 68.458206 samples/sec	 accuracy=67.125000	 loss=1.347904	 lr=0.001000
Epoch[062] Batch [0649]/[3760]	Speed: 68.376729 samples/sec	 accuracy=67.170673	 loss=1.344255	 lr=0.001000
Epoch[062] Batch [0699]/[3760]	Speed: 69.258595 samples/sec	 accuracy=67.209821	 loss=1.342800	 lr=0.001000
Epoch[062] Batch [0749]/[3760]	Speed: 69.983868 samples/sec	 accuracy=67.200000	 loss=1.345088	 lr=0.001000
Epoch[062] Batch [0799]/[3760]	Speed: 71.364788 samples/sec	 accuracy=67.126953	 loss=1.346745	 lr=0.001000
Epoch[062] Batch [0849]/[3760]	Speed: 66.595934 samples/sec	 accuracy=67.062500	 loss=1.348258	 lr=0.001000
Epoch[062] Batch [0899]/[3760]	Speed: 72.032727 samples/sec	 accuracy=67.052083	 loss=1.347982	 lr=0.001000
Epoch[062] Batch [0949]/[3760]	Speed: 70.551467 samples/sec	 accuracy=67.032895	 loss=1.348156	 lr=0.001000
Epoch[062] Batch [0999]/[3760]	Speed: 69.598734 samples/sec	 accuracy=67.023438	 loss=1.349794	 lr=0.001000
Epoch[062] Batch [1049]/[3760]	Speed: 67.924717 samples/sec	 accuracy=67.056548	 loss=1.349532	 lr=0.001000
Epoch[062] Batch [1099]/[3760]	Speed: 70.948309 samples/sec	 accuracy=67.045455	 loss=1.349283	 lr=0.001000
Epoch[062] Batch [1149]/[3760]	Speed: 69.146522 samples/sec	 accuracy=67.122283	 loss=1.346505	 lr=0.001000
Epoch[062] Batch [1199]/[3760]	Speed: 70.334429 samples/sec	 accuracy=67.093750	 loss=1.349059	 lr=0.001000
Epoch[062] Batch [1249]/[3760]	Speed: 70.012045 samples/sec	 accuracy=67.055000	 loss=1.350061	 lr=0.001000
Epoch[062] Batch [1299]/[3760]	Speed: 67.873024 samples/sec	 accuracy=67.091346	 loss=1.348305	 lr=0.001000
Epoch[062] Batch [1349]/[3760]	Speed: 70.325899 samples/sec	 accuracy=67.033565	 loss=1.350591	 lr=0.001000
Epoch[062] Batch [1399]/[3760]	Speed: 69.358332 samples/sec	 accuracy=67.011161	 loss=1.351355	 lr=0.001000
Epoch[062] Batch [1449]/[3760]	Speed: 70.872434 samples/sec	 accuracy=66.978448	 loss=1.352776	 lr=0.001000
Epoch[062] Batch [1499]/[3760]	Speed: 70.068432 samples/sec	 accuracy=66.981250	 loss=1.353873	 lr=0.001000
Epoch[062] Batch [1549]/[3760]	Speed: 68.784411 samples/sec	 accuracy=67.039315	 loss=1.350932	 lr=0.001000
Epoch[062] Batch [1599]/[3760]	Speed: 69.496928 samples/sec	 accuracy=67.036133	 loss=1.350174	 lr=0.001000
Epoch[062] Batch [1649]/[3760]	Speed: 70.101210 samples/sec	 accuracy=67.064394	 loss=1.349723	 lr=0.001000
Epoch[062] Batch [1699]/[3760]	Speed: 70.445136 samples/sec	 accuracy=67.073529	 loss=1.349097	 lr=0.001000
Epoch[062] Batch [1749]/[3760]	Speed: 68.741958 samples/sec	 accuracy=67.049107	 loss=1.349685	 lr=0.001000
Epoch[062] Batch [1799]/[3760]	Speed: 69.993459 samples/sec	 accuracy=67.032986	 loss=1.349853	 lr=0.001000
Epoch[062] Batch [1849]/[3760]	Speed: 70.539327 samples/sec	 accuracy=66.991554	 loss=1.351419	 lr=0.001000
Epoch[062] Batch [1899]/[3760]	Speed: 70.714771 samples/sec	 accuracy=66.967105	 loss=1.351667	 lr=0.001000
Epoch[062] Batch [1949]/[3760]	Speed: 67.597005 samples/sec	 accuracy=66.971955	 loss=1.351830	 lr=0.001000
Epoch[062] Batch [1999]/[3760]	Speed: 69.634574 samples/sec	 accuracy=66.955469	 loss=1.351506	 lr=0.001000
Epoch[062] Batch [2049]/[3760]	Speed: 70.318431 samples/sec	 accuracy=66.972561	 loss=1.350853	 lr=0.001000
Epoch[062] Batch [2099]/[3760]	Speed: 69.583748 samples/sec	 accuracy=66.991815	 loss=1.349843	 lr=0.001000
Epoch[062] Batch [2149]/[3760]	Speed: 70.143791 samples/sec	 accuracy=66.952035	 loss=1.351303	 lr=0.001000
Epoch[062] Batch [2199]/[3760]	Speed: 70.802597 samples/sec	 accuracy=66.926847	 loss=1.351800	 lr=0.001000
Epoch[062] Batch [2249]/[3760]	Speed: 68.514508 samples/sec	 accuracy=66.921528	 loss=1.351934	 lr=0.001000
Epoch[062] Batch [2299]/[3760]	Speed: 68.611544 samples/sec	 accuracy=66.919158	 loss=1.352855	 lr=0.001000
Epoch[062] Batch [2349]/[3760]	Speed: 72.082223 samples/sec	 accuracy=66.914894	 loss=1.352621	 lr=0.001000
Epoch[062] Batch [2399]/[3760]	Speed: 68.818355 samples/sec	 accuracy=66.882161	 loss=1.353731	 lr=0.001000
Epoch[062] Batch [2449]/[3760]	Speed: 71.028609 samples/sec	 accuracy=66.853316	 loss=1.354745	 lr=0.001000
Epoch[062] Batch [2499]/[3760]	Speed: 69.898792 samples/sec	 accuracy=66.861250	 loss=1.353708	 lr=0.001000
Epoch[062] Batch [2549]/[3760]	Speed: 69.226764 samples/sec	 accuracy=66.852328	 loss=1.353579	 lr=0.001000
Epoch[062] Batch [2599]/[3760]	Speed: 69.473396 samples/sec	 accuracy=66.846755	 loss=1.353730	 lr=0.001000
Epoch[062] Batch [2649]/[3760]	Speed: 69.759458 samples/sec	 accuracy=66.839623	 loss=1.354042	 lr=0.001000
Epoch[062] Batch [2699]/[3760]	Speed: 69.494491 samples/sec	 accuracy=66.863426	 loss=1.353153	 lr=0.001000
Epoch[062] Batch [2749]/[3760]	Speed: 68.243303 samples/sec	 accuracy=66.868750	 loss=1.352932	 lr=0.001000
Epoch[062] Batch [2799]/[3760]	Speed: 69.077922 samples/sec	 accuracy=66.853795	 loss=1.353286	 lr=0.001000
Epoch[062] Batch [2849]/[3760]	Speed: 69.630581 samples/sec	 accuracy=66.857456	 loss=1.352911	 lr=0.001000
Epoch[062] Batch [2899]/[3760]	Speed: 66.548463 samples/sec	 accuracy=66.864763	 loss=1.353101	 lr=0.001000
Epoch[062] Batch [2949]/[3760]	Speed: 70.939801 samples/sec	 accuracy=66.850106	 loss=1.353442	 lr=0.001000
Epoch[062] Batch [2999]/[3760]	Speed: 67.689037 samples/sec	 accuracy=66.843229	 loss=1.353891	 lr=0.001000
Epoch[062] Batch [3049]/[3760]	Speed: 73.278903 samples/sec	 accuracy=66.839139	 loss=1.354608	 lr=0.001000
Epoch[062] Batch [3099]/[3760]	Speed: 70.567678 samples/sec	 accuracy=66.819556	 loss=1.355620	 lr=0.001000
Epoch[062] Batch [3149]/[3760]	Speed: 68.734966 samples/sec	 accuracy=66.808532	 loss=1.356602	 lr=0.001000
Epoch[062] Batch [3199]/[3760]	Speed: 68.586532 samples/sec	 accuracy=66.828125	 loss=1.356064	 lr=0.001000
Epoch[062] Batch [3249]/[3760]	Speed: 69.978675 samples/sec	 accuracy=66.846635	 loss=1.355211	 lr=0.001000
Epoch[062] Batch [3299]/[3760]	Speed: 68.322338 samples/sec	 accuracy=66.834280	 loss=1.355978	 lr=0.001000
Epoch[062] Batch [3349]/[3760]	Speed: 71.660415 samples/sec	 accuracy=66.844683	 loss=1.355546	 lr=0.001000
Epoch[062] Batch [3399]/[3760]	Speed: 69.467195 samples/sec	 accuracy=66.841912	 loss=1.355697	 lr=0.001000
Epoch[062] Batch [3449]/[3760]	Speed: 68.735923 samples/sec	 accuracy=66.835145	 loss=1.356029	 lr=0.001000
Epoch[062] Batch [3499]/[3760]	Speed: 71.152270 samples/sec	 accuracy=66.825446	 loss=1.356275	 lr=0.001000
Epoch[062] Batch [3549]/[3760]	Speed: 65.561360 samples/sec	 accuracy=66.834507	 loss=1.356009	 lr=0.001000
Epoch[062] Batch [3599]/[3760]	Speed: 72.547354 samples/sec	 accuracy=66.831597	 loss=1.355723	 lr=0.001000
Epoch[062] Batch [3649]/[3760]	Speed: 68.695931 samples/sec	 accuracy=66.832192	 loss=1.355572	 lr=0.001000
Epoch[062] Batch [3699]/[3760]	Speed: 70.477643 samples/sec	 accuracy=66.836571	 loss=1.355369	 lr=0.001000
Epoch[062] Batch [3749]/[3760]	Speed: 77.012756 samples/sec	 accuracy=66.845417	 loss=1.355266	 lr=0.001000
Batch [0049]/[0303]: acc-top1=61.875000 acc-top5=84.437500
Batch [0099]/[0303]: acc-top1=62.171875 acc-top5=84.078125
Batch [0149]/[0303]: acc-top1=62.937500 acc-top5=84.145833
Batch [0199]/[0303]: acc-top1=63.179688 acc-top5=84.421875
Batch [0249]/[0303]: acc-top1=63.300000 acc-top5=84.656250
Batch [0299]/[0303]: acc-top1=63.088542 acc-top5=84.421875
[Epoch 062] training: accuracy=66.839262	 loss=1.355432
[Epoch 062] speed: 69 samples/sec	time cost: 3757.428137
[Epoch 062] validation: acc-top1=63.051774 acc-top5=84.431724 loss=1.594561
Epoch[063] Batch [0049]/[3760]	Speed: 48.756097 samples/sec	 accuracy=66.218750	 loss=1.381627	 lr=0.001000
Epoch[063] Batch [0099]/[3760]	Speed: 68.897540 samples/sec	 accuracy=66.593750	 loss=1.382407	 lr=0.001000
Epoch[063] Batch [0149]/[3760]	Speed: 68.214434 samples/sec	 accuracy=67.125000	 loss=1.373539	 lr=0.001000
Epoch[063] Batch [0199]/[3760]	Speed: 69.817291 samples/sec	 accuracy=67.281250	 loss=1.348061	 lr=0.001000
Epoch[063] Batch [0249]/[3760]	Speed: 70.384574 samples/sec	 accuracy=67.187500	 loss=1.355787	 lr=0.001000
Epoch[063] Batch [0299]/[3760]	Speed: 69.826759 samples/sec	 accuracy=67.213542	 loss=1.350411	 lr=0.001000
Epoch[063] Batch [0349]/[3760]	Speed: 69.222351 samples/sec	 accuracy=67.312500	 loss=1.343633	 lr=0.001000
Epoch[063] Batch [0399]/[3760]	Speed: 70.855711 samples/sec	 accuracy=67.156250	 loss=1.348236	 lr=0.001000
Epoch[063] Batch [0449]/[3760]	Speed: 67.489569 samples/sec	 accuracy=67.079861	 loss=1.350540	 lr=0.001000
Epoch[063] Batch [0499]/[3760]	Speed: 70.728883 samples/sec	 accuracy=67.046875	 loss=1.347242	 lr=0.001000
Epoch[063] Batch [0549]/[3760]	Speed: 69.944097 samples/sec	 accuracy=67.031250	 loss=1.344997	 lr=0.001000
Epoch[063] Batch [0599]/[3760]	Speed: 69.931159 samples/sec	 accuracy=67.135417	 loss=1.340366	 lr=0.001000
Epoch[063] Batch [0649]/[3760]	Speed: 68.184020 samples/sec	 accuracy=67.100962	 loss=1.342225	 lr=0.001000
Epoch[063] Batch [0699]/[3760]	Speed: 69.703845 samples/sec	 accuracy=67.107143	 loss=1.342378	 lr=0.001000
Epoch[063] Batch [0749]/[3760]	Speed: 69.818365 samples/sec	 accuracy=67.112500	 loss=1.341956	 lr=0.001000
Epoch[063] Batch [0799]/[3760]	Speed: 70.243280 samples/sec	 accuracy=67.109375	 loss=1.341923	 lr=0.001000
Epoch[063] Batch [0849]/[3760]	Speed: 69.187648 samples/sec	 accuracy=67.110294	 loss=1.342076	 lr=0.001000
Epoch[063] Batch [0899]/[3760]	Speed: 70.469113 samples/sec	 accuracy=67.092014	 loss=1.343052	 lr=0.001000
Epoch[063] Batch [0949]/[3760]	Speed: 69.218823 samples/sec	 accuracy=67.192434	 loss=1.339770	 lr=0.001000
Epoch[063] Batch [0999]/[3760]	Speed: 70.259528 samples/sec	 accuracy=67.134375	 loss=1.341703	 lr=0.001000
Epoch[063] Batch [1049]/[3760]	Speed: 70.304569 samples/sec	 accuracy=67.138393	 loss=1.342003	 lr=0.001000
Epoch[063] Batch [1099]/[3760]	Speed: 70.714199 samples/sec	 accuracy=67.157670	 loss=1.340963	 lr=0.001000
Epoch[063] Batch [1149]/[3760]	Speed: 67.036592 samples/sec	 accuracy=67.153533	 loss=1.340740	 lr=0.001000
Epoch[063] Batch [1199]/[3760]	Speed: 71.509546 samples/sec	 accuracy=67.191406	 loss=1.337991	 lr=0.001000
Epoch[063] Batch [1249]/[3760]	Speed: 68.936373 samples/sec	 accuracy=67.162500	 loss=1.338254	 lr=0.001000
Epoch[063] Batch [1299]/[3760]	Speed: 70.793165 samples/sec	 accuracy=67.152644	 loss=1.338592	 lr=0.001000
Epoch[063] Batch [1349]/[3760]	Speed: 69.849192 samples/sec	 accuracy=67.167824	 loss=1.338462	 lr=0.001000
Epoch[063] Batch [1399]/[3760]	Speed: 69.050898 samples/sec	 accuracy=67.151786	 loss=1.338297	 lr=0.001000
Epoch[063] Batch [1449]/[3760]	Speed: 71.013384 samples/sec	 accuracy=67.215517	 loss=1.336802	 lr=0.001000
Epoch[063] Batch [1499]/[3760]	Speed: 69.408536 samples/sec	 accuracy=67.198958	 loss=1.337314	 lr=0.001000
Epoch[063] Batch [1549]/[3760]	Speed: 70.339465 samples/sec	 accuracy=67.166331	 loss=1.337922	 lr=0.001000
Epoch[063] Batch [1599]/[3760]	Speed: 68.142537 samples/sec	 accuracy=67.200195	 loss=1.336180	 lr=0.001000
Epoch[063] Batch [1649]/[3760]	Speed: 69.503350 samples/sec	 accuracy=67.222538	 loss=1.336195	 lr=0.001000
Epoch[063] Batch [1699]/[3760]	Speed: 71.270434 samples/sec	 accuracy=67.219669	 loss=1.336312	 lr=0.001000
Epoch[063] Batch [1749]/[3760]	Speed: 68.356820 samples/sec	 accuracy=67.181250	 loss=1.337756	 lr=0.001000
Epoch[063] Batch [1799]/[3760]	Speed: 70.356928 samples/sec	 accuracy=67.167535	 loss=1.338365	 lr=0.001000
Epoch[063] Batch [1849]/[3760]	Speed: 70.782350 samples/sec	 accuracy=67.125000	 loss=1.339852	 lr=0.001000
Epoch[063] Batch [1899]/[3760]	Speed: 70.278621 samples/sec	 accuracy=67.115954	 loss=1.339864	 lr=0.001000
Epoch[063] Batch [1949]/[3760]	Speed: 67.836630 samples/sec	 accuracy=67.095353	 loss=1.340983	 lr=0.001000
Epoch[063] Batch [1999]/[3760]	Speed: 69.056644 samples/sec	 accuracy=67.064844	 loss=1.342271	 lr=0.001000
Epoch[063] Batch [2049]/[3760]	Speed: 71.142980 samples/sec	 accuracy=67.057165	 loss=1.343766	 lr=0.001000
Epoch[063] Batch [2099]/[3760]	Speed: 69.631419 samples/sec	 accuracy=67.055804	 loss=1.343710	 lr=0.001000
Epoch[063] Batch [2149]/[3760]	Speed: 70.283591 samples/sec	 accuracy=67.033430	 loss=1.344632	 lr=0.001000
Epoch[063] Batch [2199]/[3760]	Speed: 68.788781 samples/sec	 accuracy=67.008523	 loss=1.345764	 lr=0.001000
Epoch[063] Batch [2249]/[3760]	Speed: 70.281978 samples/sec	 accuracy=67.021528	 loss=1.345075	 lr=0.001000
Epoch[063] Batch [2299]/[3760]	Speed: 70.371490 samples/sec	 accuracy=67.025136	 loss=1.344598	 lr=0.001000
Epoch[063] Batch [2349]/[3760]	Speed: 67.114514 samples/sec	 accuracy=67.037899	 loss=1.344175	 lr=0.001000
Epoch[063] Batch [2399]/[3760]	Speed: 72.092692 samples/sec	 accuracy=66.998698	 loss=1.344856	 lr=0.001000
Epoch[063] Batch [2449]/[3760]	Speed: 69.895586 samples/sec	 accuracy=66.971939	 loss=1.345309	 lr=0.001000
Epoch[063] Batch [2499]/[3760]	Speed: 69.630522 samples/sec	 accuracy=66.965000	 loss=1.345495	 lr=0.001000
Epoch[063] Batch [2549]/[3760]	Speed: 70.224386 samples/sec	 accuracy=66.952819	 loss=1.346088	 lr=0.001000
Epoch[063] Batch [2599]/[3760]	Speed: 69.673935 samples/sec	 accuracy=66.960938	 loss=1.345579	 lr=0.001000
Epoch[063] Batch [2649]/[3760]	Speed: 70.716160 samples/sec	 accuracy=66.926297	 loss=1.346088	 lr=0.001000
Epoch[063] Batch [2699]/[3760]	Speed: 68.994886 samples/sec	 accuracy=66.940972	 loss=1.345723	 lr=0.001000
Epoch[063] Batch [2749]/[3760]	Speed: 69.920057 samples/sec	 accuracy=66.960227	 loss=1.345065	 lr=0.001000
Epoch[063] Batch [2799]/[3760]	Speed: 69.336615 samples/sec	 accuracy=66.969866	 loss=1.345298	 lr=0.001000
Epoch[063] Batch [2849]/[3760]	Speed: 71.356138 samples/sec	 accuracy=66.965461	 loss=1.345295	 lr=0.001000
Epoch[063] Batch [2899]/[3760]	Speed: 69.782438 samples/sec	 accuracy=66.945582	 loss=1.346039	 lr=0.001000
Epoch[063] Batch [2949]/[3760]	Speed: 71.237293 samples/sec	 accuracy=66.960805	 loss=1.345659	 lr=0.001000
Epoch[063] Batch [2999]/[3760]	Speed: 69.095832 samples/sec	 accuracy=66.966146	 loss=1.345744	 lr=0.001000
Epoch[063] Batch [3049]/[3760]	Speed: 69.024902 samples/sec	 accuracy=66.986168	 loss=1.344978	 lr=0.001000
Epoch[063] Batch [3099]/[3760]	Speed: 69.714559 samples/sec	 accuracy=67.007560	 loss=1.344521	 lr=0.001000
Epoch[063] Batch [3149]/[3760]	Speed: 70.155577 samples/sec	 accuracy=67.011905	 loss=1.343954	 lr=0.001000
Epoch[063] Batch [3199]/[3760]	Speed: 69.563476 samples/sec	 accuracy=67.010742	 loss=1.343930	 lr=0.001000
Epoch[063] Batch [3249]/[3760]	Speed: 68.056705 samples/sec	 accuracy=66.984135	 loss=1.344564	 lr=0.001000
Epoch[063] Batch [3299]/[3760]	Speed: 70.890921 samples/sec	 accuracy=66.983428	 loss=1.344522	 lr=0.001000
Epoch[063] Batch [3349]/[3760]	Speed: 70.125027 samples/sec	 accuracy=66.963619	 loss=1.345196	 lr=0.001000
Epoch[063] Batch [3399]/[3760]	Speed: 69.610490 samples/sec	 accuracy=66.969669	 loss=1.345118	 lr=0.001000
Epoch[063] Batch [3449]/[3760]	Speed: 69.947023 samples/sec	 accuracy=66.948822	 loss=1.345204	 lr=0.001000
Epoch[063] Batch [3499]/[3760]	Speed: 68.770500 samples/sec	 accuracy=66.946429	 loss=1.344968	 lr=0.001000
Epoch[063] Batch [3549]/[3760]	Speed: 70.277644 samples/sec	 accuracy=66.947623	 loss=1.345016	 lr=0.001000
Epoch[063] Batch [3599]/[3760]	Speed: 69.900692 samples/sec	 accuracy=66.929253	 loss=1.345304	 lr=0.001000
Epoch[063] Batch [3649]/[3760]	Speed: 70.440600 samples/sec	 accuracy=66.942637	 loss=1.345006	 lr=0.001000
Epoch[063] Batch [3699]/[3760]	Speed: 68.052457 samples/sec	 accuracy=66.949747	 loss=1.344682	 lr=0.001000
Epoch[063] Batch [3749]/[3760]	Speed: 77.641212 samples/sec	 accuracy=66.948750	 loss=1.344946	 lr=0.001000
Batch [0049]/[0303]: acc-top1=61.687500 acc-top5=83.781250
Batch [0099]/[0303]: acc-top1=61.781250 acc-top5=83.500000
Batch [0149]/[0303]: acc-top1=62.614583 acc-top5=83.968750
Batch [0199]/[0303]: acc-top1=63.031250 acc-top5=84.445312
Batch [0249]/[0303]: acc-top1=63.106250 acc-top5=84.575000
Batch [0299]/[0303]: acc-top1=62.864583 acc-top5=84.338542
[Epoch 063] training: accuracy=66.956449	 loss=1.344876
[Epoch 063] speed: 69 samples/sec	time cost: 3750.874964
[Epoch 063] validation: acc-top1=62.855817 acc-top5=84.354373 loss=1.616013
Epoch[064] Batch [0049]/[3759]	Speed: 48.819778 samples/sec	 accuracy=67.343750	 loss=1.343325	 lr=0.001000
Epoch[064] Batch [0099]/[3759]	Speed: 69.437015 samples/sec	 accuracy=67.687500	 loss=1.327776	 lr=0.001000
Epoch[064] Batch [0149]/[3759]	Speed: 68.917333 samples/sec	 accuracy=67.302083	 loss=1.343165	 lr=0.001000
Epoch[064] Batch [0199]/[3759]	Speed: 68.783191 samples/sec	 accuracy=67.593750	 loss=1.328547	 lr=0.001000
Epoch[064] Batch [0249]/[3759]	Speed: 68.547169 samples/sec	 accuracy=67.531250	 loss=1.337706	 lr=0.001000
Epoch[064] Batch [0299]/[3759]	Speed: 69.622768 samples/sec	 accuracy=67.255208	 loss=1.339107	 lr=0.001000
Epoch[064] Batch [0349]/[3759]	Speed: 70.871801 samples/sec	 accuracy=67.187500	 loss=1.333128	 lr=0.001000
Epoch[064] Batch [0399]/[3759]	Speed: 70.467647 samples/sec	 accuracy=67.078125	 loss=1.340221	 lr=0.001000
Epoch[064] Batch [0449]/[3759]	Speed: 68.031048 samples/sec	 accuracy=67.059028	 loss=1.341674	 lr=0.001000
Epoch[064] Batch [0499]/[3759]	Speed: 70.826471 samples/sec	 accuracy=67.065625	 loss=1.341545	 lr=0.001000
Epoch[064] Batch [0549]/[3759]	Speed: 67.558391 samples/sec	 accuracy=67.082386	 loss=1.343131	 lr=0.001000
Epoch[064] Batch [0599]/[3759]	Speed: 70.007032 samples/sec	 accuracy=67.101562	 loss=1.343271	 lr=0.001000
Epoch[064] Batch [0649]/[3759]	Speed: 70.137952 samples/sec	 accuracy=67.168269	 loss=1.341573	 lr=0.001000
Epoch[064] Batch [0699]/[3759]	Speed: 69.886090 samples/sec	 accuracy=67.116071	 loss=1.341265	 lr=0.001000
Epoch[064] Batch [0749]/[3759]	Speed: 69.672238 samples/sec	 accuracy=67.045833	 loss=1.340270	 lr=0.001000
Epoch[064] Batch [0799]/[3759]	Speed: 70.119136 samples/sec	 accuracy=67.130859	 loss=1.336849	 lr=0.001000
Epoch[064] Batch [0849]/[3759]	Speed: 69.920189 samples/sec	 accuracy=67.154412	 loss=1.337277	 lr=0.001000
Epoch[064] Batch [0899]/[3759]	Speed: 69.774407 samples/sec	 accuracy=67.144097	 loss=1.336238	 lr=0.001000
Epoch[064] Batch [0949]/[3759]	Speed: 68.298889 samples/sec	 accuracy=67.093750	 loss=1.338063	 lr=0.001000
Epoch[064] Batch [0999]/[3759]	Speed: 70.249107 samples/sec	 accuracy=67.103125	 loss=1.338550	 lr=0.001000
Epoch[064] Batch [1049]/[3759]	Speed: 69.867286 samples/sec	 accuracy=67.080357	 loss=1.338708	 lr=0.001000
Epoch[064] Batch [1099]/[3759]	Speed: 68.751121 samples/sec	 accuracy=67.137784	 loss=1.337268	 lr=0.001000
Epoch[064] Batch [1149]/[3759]	Speed: 70.264224 samples/sec	 accuracy=67.065217	 loss=1.339474	 lr=0.001000
Epoch[064] Batch [1199]/[3759]	Speed: 71.131258 samples/sec	 accuracy=67.065104	 loss=1.340882	 lr=0.001000
Epoch[064] Batch [1249]/[3759]	Speed: 69.602549 samples/sec	 accuracy=67.067500	 loss=1.339457	 lr=0.001000
Epoch[064] Batch [1299]/[3759]	Speed: 68.921516 samples/sec	 accuracy=67.064904	 loss=1.339088	 lr=0.001000
Epoch[064] Batch [1349]/[3759]	Speed: 70.678832 samples/sec	 accuracy=67.087963	 loss=1.337831	 lr=0.001000
Epoch[064] Batch [1399]/[3759]	Speed: 68.802397 samples/sec	 accuracy=67.075893	 loss=1.338916	 lr=0.001000
Epoch[064] Batch [1449]/[3759]	Speed: 67.816587 samples/sec	 accuracy=67.073276	 loss=1.339609	 lr=0.001000
Epoch[064] Batch [1499]/[3759]	Speed: 72.211410 samples/sec	 accuracy=67.060417	 loss=1.341747	 lr=0.001000
Epoch[064] Batch [1549]/[3759]	Speed: 69.330604 samples/sec	 accuracy=67.063508	 loss=1.341817	 lr=0.001000
Epoch[064] Batch [1599]/[3759]	Speed: 69.942930 samples/sec	 accuracy=67.080078	 loss=1.341241	 lr=0.001000
Epoch[064] Batch [1649]/[3759]	Speed: 69.701371 samples/sec	 accuracy=67.126894	 loss=1.339614	 lr=0.001000
Epoch[064] Batch [1699]/[3759]	Speed: 68.792712 samples/sec	 accuracy=67.113051	 loss=1.340621	 lr=0.001000
Epoch[064] Batch [1749]/[3759]	Speed: 69.378980 samples/sec	 accuracy=67.117857	 loss=1.339923	 lr=0.001000
Epoch[064] Batch [1799]/[3759]	Speed: 70.692438 samples/sec	 accuracy=67.103299	 loss=1.339791	 lr=0.001000
Epoch[064] Batch [1849]/[3759]	Speed: 69.783835 samples/sec	 accuracy=67.111486	 loss=1.339801	 lr=0.001000
Epoch[064] Batch [1899]/[3759]	Speed: 68.477365 samples/sec	 accuracy=67.106908	 loss=1.339510	 lr=0.001000
Epoch[064] Batch [1949]/[3759]	Speed: 70.748426 samples/sec	 accuracy=67.085737	 loss=1.339813	 lr=0.001000
Epoch[064] Batch [1999]/[3759]	Speed: 69.132249 samples/sec	 accuracy=67.083594	 loss=1.339700	 lr=0.001000
Epoch[064] Batch [2049]/[3759]	Speed: 69.367260 samples/sec	 accuracy=67.050305	 loss=1.340127	 lr=0.001000
Epoch[064] Batch [2099]/[3759]	Speed: 69.488222 samples/sec	 accuracy=67.049851	 loss=1.339489	 lr=0.001000
Epoch[064] Batch [2149]/[3759]	Speed: 69.225440 samples/sec	 accuracy=66.987645	 loss=1.342086	 lr=0.001000
Epoch[064] Batch [2199]/[3759]	Speed: 70.309726 samples/sec	 accuracy=66.958097	 loss=1.342402	 lr=0.001000
Epoch[064] Batch [2249]/[3759]	Speed: 69.017716 samples/sec	 accuracy=66.990278	 loss=1.341428	 lr=0.001000
Epoch[064] Batch [2299]/[3759]	Speed: 70.642778 samples/sec	 accuracy=67.001359	 loss=1.340960	 lr=0.001000
Epoch[064] Batch [2349]/[3759]	Speed: 69.251064 samples/sec	 accuracy=67.024601	 loss=1.340877	 lr=0.001000
Epoch[064] Batch [2399]/[3759]	Speed: 70.050631 samples/sec	 accuracy=67.029948	 loss=1.340428	 lr=0.001000
Epoch[064] Batch [2449]/[3759]	Speed: 69.875164 samples/sec	 accuracy=67.054209	 loss=1.339977	 lr=0.001000
Epoch[064] Batch [2499]/[3759]	Speed: 69.692109 samples/sec	 accuracy=67.069375	 loss=1.339058	 lr=0.001000
Epoch[064] Batch [2549]/[3759]	Speed: 71.200763 samples/sec	 accuracy=67.091912	 loss=1.338484	 lr=0.001000
Epoch[064] Batch [2599]/[3759]	Speed: 68.939853 samples/sec	 accuracy=67.084736	 loss=1.338669	 lr=0.001000
Epoch[064] Batch [2649]/[3759]	Speed: 70.107334 samples/sec	 accuracy=67.076651	 loss=1.339408	 lr=0.001000
Epoch[064] Batch [2699]/[3759]	Speed: 68.703335 samples/sec	 accuracy=67.098958	 loss=1.338849	 lr=0.001000
Epoch[064] Batch [2749]/[3759]	Speed: 67.256024 samples/sec	 accuracy=67.107955	 loss=1.338567	 lr=0.001000
Epoch[064] Batch [2799]/[3759]	Speed: 71.460774 samples/sec	 accuracy=67.112723	 loss=1.339070	 lr=0.001000
Epoch[064] Batch [2849]/[3759]	Speed: 70.707810 samples/sec	 accuracy=67.097588	 loss=1.339501	 lr=0.001000
Epoch[064] Batch [2899]/[3759]	Speed: 68.427819 samples/sec	 accuracy=67.109914	 loss=1.339260	 lr=0.001000
Epoch[064] Batch [2949]/[3759]	Speed: 70.565867 samples/sec	 accuracy=67.095869	 loss=1.339917	 lr=0.001000
Epoch[064] Batch [2999]/[3759]	Speed: 69.749595 samples/sec	 accuracy=67.081771	 loss=1.340127	 lr=0.001000
Epoch[064] Batch [3049]/[3759]	Speed: 69.629841 samples/sec	 accuracy=67.068135	 loss=1.340812	 lr=0.001000
Epoch[064] Batch [3099]/[3759]	Speed: 70.545054 samples/sec	 accuracy=67.058972	 loss=1.341302	 lr=0.001000
Epoch[064] Batch [3149]/[3759]	Speed: 67.892871 samples/sec	 accuracy=67.064980	 loss=1.341477	 lr=0.001000
Epoch[064] Batch [3199]/[3759]	Speed: 71.204978 samples/sec	 accuracy=67.051270	 loss=1.341879	 lr=0.001000
Epoch[064] Batch [3249]/[3759]	Speed: 69.097194 samples/sec	 accuracy=67.057212	 loss=1.341001	 lr=0.001000
Epoch[064] Batch [3299]/[3759]	Speed: 69.472505 samples/sec	 accuracy=67.044034	 loss=1.340959	 lr=0.001000
Epoch[064] Batch [3349]/[3759]	Speed: 70.865873 samples/sec	 accuracy=67.050373	 loss=1.340705	 lr=0.001000
Epoch[064] Batch [3399]/[3759]	Speed: 68.599736 samples/sec	 accuracy=67.031710	 loss=1.341307	 lr=0.001000
Epoch[064] Batch [3449]/[3759]	Speed: 70.779481 samples/sec	 accuracy=67.021286	 loss=1.341523	 lr=0.001000
Epoch[064] Batch [3499]/[3759]	Speed: 68.786641 samples/sec	 accuracy=67.008482	 loss=1.342288	 lr=0.001000
Epoch[064] Batch [3549]/[3759]	Speed: 65.817945 samples/sec	 accuracy=67.003081	 loss=1.342990	 lr=0.001000
Epoch[064] Batch [3599]/[3759]	Speed: 72.112498 samples/sec	 accuracy=67.013021	 loss=1.343423	 lr=0.001000
Epoch[064] Batch [3649]/[3759]	Speed: 70.384808 samples/sec	 accuracy=67.015411	 loss=1.343405	 lr=0.001000
Epoch[064] Batch [3699]/[3759]	Speed: 70.774159 samples/sec	 accuracy=67.018581	 loss=1.343370	 lr=0.001000
Epoch[064] Batch [3749]/[3759]	Speed: 77.463549 samples/sec	 accuracy=67.025000	 loss=1.343713	 lr=0.001000
Batch [0049]/[0303]: acc-top1=62.031250 acc-top5=84.187500
Batch [0099]/[0303]: acc-top1=62.000000 acc-top5=83.546875
Batch [0149]/[0303]: acc-top1=62.687500 acc-top5=84.000000
Batch [0199]/[0303]: acc-top1=63.132812 acc-top5=84.343750
Batch [0249]/[0303]: acc-top1=63.212500 acc-top5=84.612500
Batch [0299]/[0303]: acc-top1=63.010417 acc-top5=84.484375
[Epoch 064] training: accuracy=67.019154	 loss=1.343893
[Epoch 064] speed: 69 samples/sec	time cost: 3754.396767
[Epoch 064] validation: acc-top1=62.974422 acc-top5=84.483292 loss=1.602141
Epoch[065] Batch [0049]/[3760]	Speed: 46.112740 samples/sec	 accuracy=67.875000	 loss=1.309230	 lr=0.001000
Epoch[065] Batch [0099]/[3760]	Speed: 71.275008 samples/sec	 accuracy=67.656250	 loss=1.315710	 lr=0.001000
Epoch[065] Batch [0149]/[3760]	Speed: 70.069867 samples/sec	 accuracy=67.510417	 loss=1.325287	 lr=0.001000
Epoch[065] Batch [0199]/[3760]	Speed: 69.374963 samples/sec	 accuracy=67.687500	 loss=1.330394	 lr=0.001000
Epoch[065] Batch [0249]/[3760]	Speed: 69.004488 samples/sec	 accuracy=67.375000	 loss=1.338544	 lr=0.001000
Epoch[065] Batch [0299]/[3760]	Speed: 69.743907 samples/sec	 accuracy=67.276042	 loss=1.339828	 lr=0.001000
Epoch[065] Batch [0349]/[3760]	Speed: 70.336540 samples/sec	 accuracy=67.133929	 loss=1.345905	 lr=0.001000
Epoch[065] Batch [0399]/[3760]	Speed: 67.678606 samples/sec	 accuracy=67.230469	 loss=1.341577	 lr=0.001000
Epoch[065] Batch [0449]/[3760]	Speed: 72.103279 samples/sec	 accuracy=67.274306	 loss=1.337412	 lr=0.001000
Epoch[065] Batch [0499]/[3760]	Speed: 66.451800 samples/sec	 accuracy=67.359375	 loss=1.335486	 lr=0.001000
Epoch[065] Batch [0549]/[3760]	Speed: 72.039132 samples/sec	 accuracy=67.397727	 loss=1.333250	 lr=0.001000
Epoch[065] Batch [0599]/[3760]	Speed: 70.373304 samples/sec	 accuracy=67.390625	 loss=1.334651	 lr=0.001000
Epoch[065] Batch [0649]/[3760]	Speed: 69.808713 samples/sec	 accuracy=67.319712	 loss=1.337667	 lr=0.001000
Epoch[065] Batch [0699]/[3760]	Speed: 68.850918 samples/sec	 accuracy=67.265625	 loss=1.337765	 lr=0.001000
Epoch[065] Batch [0749]/[3760]	Speed: 70.108624 samples/sec	 accuracy=67.210417	 loss=1.341007	 lr=0.001000
Epoch[065] Batch [0799]/[3760]	Speed: 69.097236 samples/sec	 accuracy=67.230469	 loss=1.340618	 lr=0.001000
Epoch[065] Batch [0849]/[3760]	Speed: 70.193435 samples/sec	 accuracy=67.229779	 loss=1.340288	 lr=0.001000
Epoch[065] Batch [0899]/[3760]	Speed: 67.654730 samples/sec	 accuracy=67.234375	 loss=1.339379	 lr=0.001000
Epoch[065] Batch [0949]/[3760]	Speed: 70.469370 samples/sec	 accuracy=67.271382	 loss=1.338820	 lr=0.001000
Epoch[065] Batch [0999]/[3760]	Speed: 71.143540 samples/sec	 accuracy=67.234375	 loss=1.339860	 lr=0.001000
Epoch[065] Batch [1049]/[3760]	Speed: 70.002307 samples/sec	 accuracy=67.238095	 loss=1.339803	 lr=0.001000
Epoch[065] Batch [1099]/[3760]	Speed: 68.841039 samples/sec	 accuracy=67.261364	 loss=1.337801	 lr=0.001000
Epoch[065] Batch [1149]/[3760]	Speed: 68.864964 samples/sec	 accuracy=67.324728	 loss=1.336642	 lr=0.001000
Epoch[065] Batch [1199]/[3760]	Speed: 70.225263 samples/sec	 accuracy=67.265625	 loss=1.337734	 lr=0.001000
Epoch[065] Batch [1249]/[3760]	Speed: 70.665156 samples/sec	 accuracy=67.245000	 loss=1.337361	 lr=0.001000
Epoch[065] Batch [1299]/[3760]	Speed: 69.211246 samples/sec	 accuracy=67.262019	 loss=1.336930	 lr=0.001000
Epoch[065] Batch [1349]/[3760]	Speed: 68.018949 samples/sec	 accuracy=67.246528	 loss=1.337417	 lr=0.001000
Epoch[065] Batch [1399]/[3760]	Speed: 68.366733 samples/sec	 accuracy=67.237723	 loss=1.338556	 lr=0.001000
Epoch[065] Batch [1449]/[3760]	Speed: 72.197239 samples/sec	 accuracy=67.218750	 loss=1.340007	 lr=0.001000
Epoch[065] Batch [1499]/[3760]	Speed: 70.974621 samples/sec	 accuracy=67.231250	 loss=1.338834	 lr=0.001000
Epoch[065] Batch [1549]/[3760]	Speed: 69.568533 samples/sec	 accuracy=67.171371	 loss=1.340549	 lr=0.001000
Epoch[065] Batch [1599]/[3760]	Speed: 67.949375 samples/sec	 accuracy=67.197266	 loss=1.339343	 lr=0.001000
Epoch[065] Batch [1649]/[3760]	Speed: 71.349405 samples/sec	 accuracy=67.199811	 loss=1.338673	 lr=0.001000
Epoch[065] Batch [1699]/[3760]	Speed: 69.220494 samples/sec	 accuracy=67.203125	 loss=1.340104	 lr=0.001000
Epoch[065] Batch [1749]/[3760]	Speed: 69.336538 samples/sec	 accuracy=67.246429	 loss=1.337716	 lr=0.001000
Epoch[065] Batch [1799]/[3760]	Speed: 70.045359 samples/sec	 accuracy=67.255208	 loss=1.336659	 lr=0.001000
Epoch[065] Batch [1849]/[3760]	Speed: 69.339890 samples/sec	 accuracy=67.275338	 loss=1.335413	 lr=0.001000
Epoch[065] Batch [1899]/[3760]	Speed: 70.248201 samples/sec	 accuracy=67.245888	 loss=1.336144	 lr=0.001000
Epoch[065] Batch [1949]/[3760]	Speed: 69.256344 samples/sec	 accuracy=67.266026	 loss=1.335657	 lr=0.001000
Epoch[065] Batch [1999]/[3760]	Speed: 68.933023 samples/sec	 accuracy=67.304688	 loss=1.334356	 lr=0.001000
Epoch[065] Batch [2049]/[3760]	Speed: 69.420420 samples/sec	 accuracy=67.269055	 loss=1.335576	 lr=0.001000
Epoch[065] Batch [2099]/[3760]	Speed: 70.324357 samples/sec	 accuracy=67.285714	 loss=1.335304	 lr=0.001000
Epoch[065] Batch [2149]/[3760]	Speed: 69.814430 samples/sec	 accuracy=67.267442	 loss=1.335075	 lr=0.001000
Epoch[065] Batch [2199]/[3760]	Speed: 68.269974 samples/sec	 accuracy=67.276278	 loss=1.334352	 lr=0.001000
Epoch[065] Batch [2249]/[3760]	Speed: 70.968572 samples/sec	 accuracy=67.248611	 loss=1.335808	 lr=0.001000
Epoch[065] Batch [2299]/[3760]	Speed: 69.456299 samples/sec	 accuracy=67.254755	 loss=1.335522	 lr=0.001000
Epoch[065] Batch [2349]/[3760]	Speed: 70.303778 samples/sec	 accuracy=67.257314	 loss=1.335499	 lr=0.001000
Epoch[065] Batch [2399]/[3760]	Speed: 70.128700 samples/sec	 accuracy=67.241536	 loss=1.335166	 lr=0.001000
Epoch[065] Batch [2449]/[3760]	Speed: 69.409630 samples/sec	 accuracy=67.242347	 loss=1.334718	 lr=0.001000
Epoch[065] Batch [2499]/[3760]	Speed: 70.057151 samples/sec	 accuracy=67.228750	 loss=1.334699	 lr=0.001000
Epoch[065] Batch [2549]/[3760]	Speed: 69.349295 samples/sec	 accuracy=67.199142	 loss=1.336144	 lr=0.001000
Epoch[065] Batch [2599]/[3760]	Speed: 70.033276 samples/sec	 accuracy=67.214543	 loss=1.336533	 lr=0.001000
Epoch[065] Batch [2649]/[3760]	Speed: 69.250114 samples/sec	 accuracy=67.198113	 loss=1.337107	 lr=0.001000
Epoch[065] Batch [2699]/[3760]	Speed: 71.527274 samples/sec	 accuracy=67.196759	 loss=1.336338	 lr=0.001000
Epoch[065] Batch [2749]/[3760]	Speed: 68.630628 samples/sec	 accuracy=67.190909	 loss=1.336712	 lr=0.001000
Epoch[065] Batch [2799]/[3760]	Speed: 68.993383 samples/sec	 accuracy=67.172433	 loss=1.337282	 lr=0.001000
Epoch[065] Batch [2849]/[3760]	Speed: 69.772014 samples/sec	 accuracy=67.202303	 loss=1.336291	 lr=0.001000
Epoch[065] Batch [2899]/[3760]	Speed: 70.088940 samples/sec	 accuracy=67.184806	 loss=1.336756	 lr=0.001000
Epoch[065] Batch [2949]/[3760]	Speed: 70.506468 samples/sec	 accuracy=67.170551	 loss=1.337291	 lr=0.001000
Epoch[065] Batch [2999]/[3760]	Speed: 68.634229 samples/sec	 accuracy=67.169792	 loss=1.337382	 lr=0.001000
Epoch[065] Batch [3049]/[3760]	Speed: 69.917945 samples/sec	 accuracy=67.158299	 loss=1.337388	 lr=0.001000
Epoch[065] Batch [3099]/[3760]	Speed: 64.951756 samples/sec	 accuracy=67.179940	 loss=1.337130	 lr=0.001000
Epoch[065] Batch [3149]/[3760]	Speed: 72.256094 samples/sec	 accuracy=67.180556	 loss=1.337411	 lr=0.001000
Epoch[065] Batch [3199]/[3760]	Speed: 71.239178 samples/sec	 accuracy=67.180664	 loss=1.337488	 lr=0.001000
Epoch[065] Batch [3249]/[3760]	Speed: 70.498702 samples/sec	 accuracy=67.159135	 loss=1.338028	 lr=0.001000
Epoch[065] Batch [3299]/[3760]	Speed: 68.889700 samples/sec	 accuracy=67.156723	 loss=1.338379	 lr=0.001000
Epoch[065] Batch [3349]/[3760]	Speed: 70.416729 samples/sec	 accuracy=67.139459	 loss=1.338915	 lr=0.001000
Epoch[065] Batch [3399]/[3760]	Speed: 69.114203 samples/sec	 accuracy=67.129136	 loss=1.339582	 lr=0.001000
Epoch[065] Batch [3449]/[3760]	Speed: 68.116850 samples/sec	 accuracy=67.138134	 loss=1.339415	 lr=0.001000
Epoch[065] Batch [3499]/[3760]	Speed: 69.689381 samples/sec	 accuracy=67.141964	 loss=1.339569	 lr=0.001000
Epoch[065] Batch [3549]/[3760]	Speed: 70.176338 samples/sec	 accuracy=67.158011	 loss=1.339331	 lr=0.001000
Epoch[065] Batch [3599]/[3760]	Speed: 67.755907 samples/sec	 accuracy=67.159288	 loss=1.339368	 lr=0.001000
Epoch[065] Batch [3649]/[3760]	Speed: 71.857687 samples/sec	 accuracy=67.122860	 loss=1.340210	 lr=0.001000
Epoch[065] Batch [3699]/[3760]	Speed: 69.896563 samples/sec	 accuracy=67.102618	 loss=1.340493	 lr=0.001000
Epoch[065] Batch [3749]/[3760]	Speed: 75.956104 samples/sec	 accuracy=67.102083	 loss=1.340517	 lr=0.001000
Batch [0049]/[0303]: acc-top1=62.531250 acc-top5=83.781250
Batch [0099]/[0303]: acc-top1=62.031250 acc-top5=83.500000
Batch [0149]/[0303]: acc-top1=62.906250 acc-top5=83.979167
Batch [0199]/[0303]: acc-top1=63.164062 acc-top5=84.335938
Batch [0249]/[0303]: acc-top1=63.275000 acc-top5=84.531250
Batch [0299]/[0303]: acc-top1=62.994792 acc-top5=84.338542
[Epoch 065] training: accuracy=67.102726	 loss=1.340655
[Epoch 065] speed: 69 samples/sec	time cost: 3757.273819
[Epoch 065] validation: acc-top1=62.953795 acc-top5=84.359530 loss=1.606797
Epoch[066] Batch [0049]/[3759]	Speed: 48.631939 samples/sec	 accuracy=66.531250	 loss=1.366056	 lr=0.001000
Epoch[066] Batch [0099]/[3759]	Speed: 68.433100 samples/sec	 accuracy=67.062500	 loss=1.344378	 lr=0.001000
Epoch[066] Batch [0149]/[3759]	Speed: 66.848198 samples/sec	 accuracy=66.875000	 loss=1.353218	 lr=0.001000
Epoch[066] Batch [0199]/[3759]	Speed: 71.230590 samples/sec	 accuracy=67.156250	 loss=1.342914	 lr=0.001000
Epoch[066] Batch [0249]/[3759]	Speed: 69.721908 samples/sec	 accuracy=67.100000	 loss=1.345503	 lr=0.001000
Epoch[066] Batch [0299]/[3759]	Speed: 70.359313 samples/sec	 accuracy=66.875000	 loss=1.347515	 lr=0.001000
Epoch[066] Batch [0349]/[3759]	Speed: 69.956093 samples/sec	 accuracy=66.812500	 loss=1.346892	 lr=0.001000
Epoch[066] Batch [0399]/[3759]	Speed: 69.581807 samples/sec	 accuracy=66.816406	 loss=1.343153	 lr=0.001000
Epoch[066] Batch [0449]/[3759]	Speed: 69.882200 samples/sec	 accuracy=66.940972	 loss=1.340547	 lr=0.001000
Epoch[066] Batch [0499]/[3759]	Speed: 69.167662 samples/sec	 accuracy=66.871875	 loss=1.341666	 lr=0.001000
Epoch[066] Batch [0549]/[3759]	Speed: 70.678446 samples/sec	 accuracy=67.022727	 loss=1.336813	 lr=0.001000
Epoch[066] Batch [0599]/[3759]	Speed: 68.789919 samples/sec	 accuracy=67.218750	 loss=1.331124	 lr=0.001000
Epoch[066] Batch [0649]/[3759]	Speed: 70.924005 samples/sec	 accuracy=67.223558	 loss=1.331024	 lr=0.001000
Epoch[066] Batch [0699]/[3759]	Speed: 69.287999 samples/sec	 accuracy=67.303571	 loss=1.329590	 lr=0.001000
Epoch[066] Batch [0749]/[3759]	Speed: 70.289488 samples/sec	 accuracy=67.291667	 loss=1.328528	 lr=0.001000
Epoch[066] Batch [0799]/[3759]	Speed: 68.866788 samples/sec	 accuracy=67.212891	 loss=1.329116	 lr=0.001000
Epoch[066] Batch [0849]/[3759]	Speed: 69.186303 samples/sec	 accuracy=67.196691	 loss=1.329062	 lr=0.001000
Epoch[066] Batch [0899]/[3759]	Speed: 71.133739 samples/sec	 accuracy=67.230903	 loss=1.326495	 lr=0.001000
Epoch[066] Batch [0949]/[3759]	Speed: 67.372031 samples/sec	 accuracy=67.187500	 loss=1.328094	 lr=0.001000
Epoch[066] Batch [0999]/[3759]	Speed: 71.218994 samples/sec	 accuracy=67.109375	 loss=1.330714	 lr=0.001000
Epoch[066] Batch [1049]/[3759]	Speed: 70.524764 samples/sec	 accuracy=67.153274	 loss=1.330730	 lr=0.001000
Epoch[066] Batch [1099]/[3759]	Speed: 70.100804 samples/sec	 accuracy=67.154830	 loss=1.331842	 lr=0.001000
Epoch[066] Batch [1149]/[3759]	Speed: 67.926785 samples/sec	 accuracy=67.100543	 loss=1.334066	 lr=0.001000
Epoch[066] Batch [1199]/[3759]	Speed: 70.048502 samples/sec	 accuracy=67.138021	 loss=1.332473	 lr=0.001000
Epoch[066] Batch [1249]/[3759]	Speed: 69.901880 samples/sec	 accuracy=67.128750	 loss=1.331860	 lr=0.001000
Epoch[066] Batch [1299]/[3759]	Speed: 71.024405 samples/sec	 accuracy=67.133413	 loss=1.331869	 lr=0.001000
Epoch[066] Batch [1349]/[3759]	Speed: 68.483980 samples/sec	 accuracy=67.173611	 loss=1.330625	 lr=0.001000
Epoch[066] Batch [1399]/[3759]	Speed: 69.093047 samples/sec	 accuracy=67.194196	 loss=1.329902	 lr=0.001000
Epoch[066] Batch [1449]/[3759]	Speed: 70.506051 samples/sec	 accuracy=67.174569	 loss=1.331103	 lr=0.001000
Epoch[066] Batch [1499]/[3759]	Speed: 70.302534 samples/sec	 accuracy=67.167708	 loss=1.331773	 lr=0.001000
Epoch[066] Batch [1549]/[3759]	Speed: 70.129628 samples/sec	 accuracy=67.235887	 loss=1.329740	 lr=0.001000
Epoch[066] Batch [1599]/[3759]	Speed: 70.020321 samples/sec	 accuracy=67.251953	 loss=1.330271	 lr=0.001000
Epoch[066] Batch [1649]/[3759]	Speed: 68.694047 samples/sec	 accuracy=67.299242	 loss=1.328470	 lr=0.001000
Epoch[066] Batch [1699]/[3759]	Speed: 69.681647 samples/sec	 accuracy=67.340993	 loss=1.327505	 lr=0.001000
Epoch[066] Batch [1749]/[3759]	Speed: 68.510059 samples/sec	 accuracy=67.294643	 loss=1.328181	 lr=0.001000
Epoch[066] Batch [1799]/[3759]	Speed: 69.357452 samples/sec	 accuracy=67.241319	 loss=1.330019	 lr=0.001000
Epoch[066] Batch [1849]/[3759]	Speed: 69.438851 samples/sec	 accuracy=67.265203	 loss=1.330148	 lr=0.001000
Epoch[066] Batch [1899]/[3759]	Speed: 70.730976 samples/sec	 accuracy=67.286184	 loss=1.330030	 lr=0.001000
Epoch[066] Batch [1949]/[3759]	Speed: 69.618059 samples/sec	 accuracy=67.302083	 loss=1.330419	 lr=0.001000
Epoch[066] Batch [1999]/[3759]	Speed: 69.769806 samples/sec	 accuracy=67.300000	 loss=1.329599	 lr=0.001000
Epoch[066] Batch [2049]/[3759]	Speed: 68.067801 samples/sec	 accuracy=67.326220	 loss=1.329002	 lr=0.001000
Epoch[066] Batch [2099]/[3759]	Speed: 70.640482 samples/sec	 accuracy=67.366815	 loss=1.328273	 lr=0.001000
Epoch[066] Batch [2149]/[3759]	Speed: 70.234648 samples/sec	 accuracy=67.333576	 loss=1.329836	 lr=0.001000
Epoch[066] Batch [2199]/[3759]	Speed: 69.279144 samples/sec	 accuracy=67.337358	 loss=1.329840	 lr=0.001000
Epoch[066] Batch [2249]/[3759]	Speed: 70.430361 samples/sec	 accuracy=67.321528	 loss=1.330153	 lr=0.001000
Epoch[066] Batch [2299]/[3759]	Speed: 69.673831 samples/sec	 accuracy=67.317935	 loss=1.330874	 lr=0.001000
Epoch[066] Batch [2349]/[3759]	Speed: 69.148192 samples/sec	 accuracy=67.325133	 loss=1.330584	 lr=0.001000
Epoch[066] Batch [2399]/[3759]	Speed: 69.783628 samples/sec	 accuracy=67.319661	 loss=1.330522	 lr=0.001000
Epoch[066] Batch [2449]/[3759]	Speed: 69.883361 samples/sec	 accuracy=67.313138	 loss=1.331005	 lr=0.001000
Epoch[066] Batch [2499]/[3759]	Speed: 68.978920 samples/sec	 accuracy=67.301250	 loss=1.331432	 lr=0.001000
Epoch[066] Batch [2549]/[3759]	Speed: 69.419167 samples/sec	 accuracy=67.266544	 loss=1.332597	 lr=0.001000
Epoch[066] Batch [2599]/[3759]	Speed: 70.828397 samples/sec	 accuracy=67.302885	 loss=1.331247	 lr=0.001000
Epoch[066] Batch [2649]/[3759]	Speed: 69.819084 samples/sec	 accuracy=67.289505	 loss=1.331807	 lr=0.001000
Epoch[066] Batch [2699]/[3759]	Speed: 70.745315 samples/sec	 accuracy=67.310764	 loss=1.331233	 lr=0.001000
Epoch[066] Batch [2749]/[3759]	Speed: 69.875518 samples/sec	 accuracy=67.314205	 loss=1.331103	 lr=0.001000
Epoch[066] Batch [2799]/[3759]	Speed: 68.512002 samples/sec	 accuracy=67.335938	 loss=1.330336	 lr=0.001000
Epoch[066] Batch [2849]/[3759]	Speed: 70.264508 samples/sec	 accuracy=67.347039	 loss=1.330405	 lr=0.001000
Epoch[066] Batch [2899]/[3759]	Speed: 70.654351 samples/sec	 accuracy=67.343750	 loss=1.330321	 lr=0.001000
Epoch[066] Batch [2949]/[3759]	Speed: 68.668237 samples/sec	 accuracy=67.342691	 loss=1.330652	 lr=0.001000
Epoch[066] Batch [2999]/[3759]	Speed: 69.561858 samples/sec	 accuracy=67.372396	 loss=1.329230	 lr=0.001000
Epoch[066] Batch [3049]/[3759]	Speed: 70.527649 samples/sec	 accuracy=67.356045	 loss=1.330283	 lr=0.001000
Epoch[066] Batch [3099]/[3759]	Speed: 69.118410 samples/sec	 accuracy=67.361895	 loss=1.330350	 lr=0.001000
Epoch[066] Batch [3149]/[3759]	Speed: 68.642152 samples/sec	 accuracy=67.349702	 loss=1.330525	 lr=0.001000
Epoch[066] Batch [3199]/[3759]	Speed: 70.525031 samples/sec	 accuracy=67.327637	 loss=1.330837	 lr=0.001000
Epoch[066] Batch [3249]/[3759]	Speed: 70.252873 samples/sec	 accuracy=67.331250	 loss=1.330902	 lr=0.001000
Epoch[066] Batch [3299]/[3759]	Speed: 69.904541 samples/sec	 accuracy=67.347538	 loss=1.330780	 lr=0.001000
Epoch[066] Batch [3349]/[3759]	Speed: 70.362195 samples/sec	 accuracy=67.326026	 loss=1.330991	 lr=0.001000
Epoch[066] Batch [3399]/[3759]	Speed: 69.617171 samples/sec	 accuracy=67.309743	 loss=1.331758	 lr=0.001000
Epoch[066] Batch [3449]/[3759]	Speed: 69.925548 samples/sec	 accuracy=67.310236	 loss=1.331915	 lr=0.001000
Epoch[066] Batch [3499]/[3759]	Speed: 69.263432 samples/sec	 accuracy=67.300000	 loss=1.332100	 lr=0.001000
Epoch[066] Batch [3549]/[3759]	Speed: 69.866037 samples/sec	 accuracy=67.298415	 loss=1.332453	 lr=0.001000
Epoch[066] Batch [3599]/[3759]	Speed: 69.629592 samples/sec	 accuracy=67.281684	 loss=1.333098	 lr=0.001000
Epoch[066] Batch [3649]/[3759]	Speed: 70.734287 samples/sec	 accuracy=67.268408	 loss=1.333871	 lr=0.001000
Epoch[066] Batch [3699]/[3759]	Speed: 69.098004 samples/sec	 accuracy=67.272804	 loss=1.333911	 lr=0.001000
Epoch[066] Batch [3749]/[3759]	Speed: 78.570006 samples/sec	 accuracy=67.277917	 loss=1.333770	 lr=0.001000
Batch [0049]/[0303]: acc-top1=61.687500 acc-top5=83.750000
Batch [0099]/[0303]: acc-top1=61.937500 acc-top5=83.515625
Batch [0149]/[0303]: acc-top1=62.645833 acc-top5=84.000000
Batch [0199]/[0303]: acc-top1=63.265625 acc-top5=84.437500
Batch [0249]/[0303]: acc-top1=63.250000 acc-top5=84.506250
Batch [0299]/[0303]: acc-top1=63.104167 acc-top5=84.411458
[Epoch 066] training: accuracy=67.284351	 loss=1.333548
[Epoch 066] speed: 69 samples/sec	time cost: 3752.497294
[Epoch 066] validation: acc-top1=63.077558 acc-top5=84.421411 loss=1.600982
Epoch[067] Batch [0049]/[3760]	Speed: 47.317349 samples/sec	 accuracy=67.875000	 loss=1.318365	 lr=0.001000
Epoch[067] Batch [0099]/[3760]	Speed: 69.557706 samples/sec	 accuracy=67.515625	 loss=1.322251	 lr=0.001000
Epoch[067] Batch [0149]/[3760]	Speed: 69.066996 samples/sec	 accuracy=67.197917	 loss=1.318128	 lr=0.001000
Epoch[067] Batch [0199]/[3760]	Speed: 69.984520 samples/sec	 accuracy=67.250000	 loss=1.322309	 lr=0.001000
Epoch[067] Batch [0249]/[3760]	Speed: 69.629335 samples/sec	 accuracy=67.387500	 loss=1.312700	 lr=0.001000
Epoch[067] Batch [0299]/[3760]	Speed: 70.342413 samples/sec	 accuracy=67.604167	 loss=1.308787	 lr=0.001000
Epoch[067] Batch [0349]/[3760]	Speed: 68.858036 samples/sec	 accuracy=67.575893	 loss=1.306207	 lr=0.001000
Epoch[067] Batch [0399]/[3760]	Speed: 70.300123 samples/sec	 accuracy=67.554688	 loss=1.308674	 lr=0.001000
Epoch[067] Batch [0449]/[3760]	Speed: 68.238878 samples/sec	 accuracy=67.475694	 loss=1.313537	 lr=0.001000
Epoch[067] Batch [0499]/[3760]	Speed: 70.383644 samples/sec	 accuracy=67.534375	 loss=1.312620	 lr=0.001000
Epoch[067] Batch [0549]/[3760]	Speed: 70.243422 samples/sec	 accuracy=67.420455	 loss=1.318514	 lr=0.001000
Epoch[067] Batch [0599]/[3760]	Speed: 70.075259 samples/sec	 accuracy=67.429688	 loss=1.320434	 lr=0.001000
Epoch[067] Batch [0649]/[3760]	Speed: 70.935328 samples/sec	 accuracy=67.454327	 loss=1.322299	 lr=0.001000
Epoch[067] Batch [0699]/[3760]	Speed: 68.633874 samples/sec	 accuracy=67.555804	 loss=1.318268	 lr=0.001000
Epoch[067] Batch [0749]/[3760]	Speed: 69.899541 samples/sec	 accuracy=67.508333	 loss=1.321007	 lr=0.001000
Epoch[067] Batch [0799]/[3760]	Speed: 69.191985 samples/sec	 accuracy=67.484375	 loss=1.319218	 lr=0.001000
Epoch[067] Batch [0849]/[3760]	Speed: 67.760292 samples/sec	 accuracy=67.419118	 loss=1.322197	 lr=0.001000
Epoch[067] Batch [0899]/[3760]	Speed: 71.899249 samples/sec	 accuracy=67.395833	 loss=1.322983	 lr=0.001000
Epoch[067] Batch [0949]/[3760]	Speed: 69.422898 samples/sec	 accuracy=67.370066	 loss=1.321886	 lr=0.001000
Epoch[067] Batch [0999]/[3760]	Speed: 70.214638 samples/sec	 accuracy=67.468750	 loss=1.319841	 lr=0.001000
Epoch[067] Batch [1049]/[3760]	Speed: 68.005635 samples/sec	 accuracy=67.525298	 loss=1.317894	 lr=0.001000
Epoch[067] Batch [1099]/[3760]	Speed: 70.272684 samples/sec	 accuracy=67.500000	 loss=1.318381	 lr=0.001000
Epoch[067] Batch [1149]/[3760]	Speed: 68.964020 samples/sec	 accuracy=67.474185	 loss=1.319668	 lr=0.001000
Epoch[067] Batch [1199]/[3760]	Speed: 70.303471 samples/sec	 accuracy=67.501302	 loss=1.319626	 lr=0.001000
Epoch[067] Batch [1249]/[3760]	Speed: 68.178717 samples/sec	 accuracy=67.518750	 loss=1.319661	 lr=0.001000
Epoch[067] Batch [1299]/[3760]	Speed: 70.751914 samples/sec	 accuracy=67.504808	 loss=1.320387	 lr=0.001000
Epoch[067] Batch [1349]/[3760]	Speed: 70.540977 samples/sec	 accuracy=67.468750	 loss=1.322025	 lr=0.001000
Epoch[067] Batch [1399]/[3760]	Speed: 69.065715 samples/sec	 accuracy=67.500000	 loss=1.321250	 lr=0.001000
Epoch[067] Batch [1449]/[3760]	Speed: 67.698831 samples/sec	 accuracy=67.507543	 loss=1.321055	 lr=0.001000
Epoch[067] Batch [1499]/[3760]	Speed: 71.423631 samples/sec	 accuracy=67.473958	 loss=1.321693	 lr=0.001000
Epoch[067] Batch [1549]/[3760]	Speed: 70.005097 samples/sec	 accuracy=67.498992	 loss=1.321071	 lr=0.001000
Epoch[067] Batch [1599]/[3760]	Speed: 69.513266 samples/sec	 accuracy=67.487305	 loss=1.321530	 lr=0.001000
Epoch[067] Batch [1649]/[3760]	Speed: 69.391089 samples/sec	 accuracy=67.504735	 loss=1.320293	 lr=0.001000
Epoch[067] Batch [1699]/[3760]	Speed: 68.315057 samples/sec	 accuracy=67.491728	 loss=1.320247	 lr=0.001000
Epoch[067] Batch [1749]/[3760]	Speed: 70.669056 samples/sec	 accuracy=67.466964	 loss=1.321125	 lr=0.001000
Epoch[067] Batch [1799]/[3760]	Speed: 70.722630 samples/sec	 accuracy=67.433160	 loss=1.321789	 lr=0.001000
Epoch[067] Batch [1849]/[3760]	Speed: 70.324572 samples/sec	 accuracy=67.412162	 loss=1.322775	 lr=0.001000
Epoch[067] Batch [1899]/[3760]	Speed: 69.084114 samples/sec	 accuracy=67.433388	 loss=1.322630	 lr=0.001000
Epoch[067] Batch [1949]/[3760]	Speed: 67.258526 samples/sec	 accuracy=67.416667	 loss=1.323689	 lr=0.001000
Epoch[067] Batch [1999]/[3760]	Speed: 70.376382 samples/sec	 accuracy=67.457031	 loss=1.323084	 lr=0.001000
Epoch[067] Batch [2049]/[3760]	Speed: 71.319875 samples/sec	 accuracy=67.447409	 loss=1.323887	 lr=0.001000
Epoch[067] Batch [2099]/[3760]	Speed: 70.800020 samples/sec	 accuracy=67.452381	 loss=1.323754	 lr=0.001000
Epoch[067] Batch [2149]/[3760]	Speed: 69.549597 samples/sec	 accuracy=67.459302	 loss=1.323360	 lr=0.001000
Epoch[067] Batch [2199]/[3760]	Speed: 68.440998 samples/sec	 accuracy=67.461648	 loss=1.322780	 lr=0.001000
Epoch[067] Batch [2249]/[3760]	Speed: 70.269151 samples/sec	 accuracy=67.482639	 loss=1.321918	 lr=0.001000
Epoch[067] Batch [2299]/[3760]	Speed: 69.249464 samples/sec	 accuracy=67.523777	 loss=1.320671	 lr=0.001000
Epoch[067] Batch [2349]/[3760]	Speed: 69.988236 samples/sec	 accuracy=67.484043	 loss=1.321892	 lr=0.001000
Epoch[067] Batch [2399]/[3760]	Speed: 68.137483 samples/sec	 accuracy=67.442708	 loss=1.323019	 lr=0.001000
Epoch[067] Batch [2449]/[3760]	Speed: 71.630328 samples/sec	 accuracy=67.459821	 loss=1.322467	 lr=0.001000
Epoch[067] Batch [2499]/[3760]	Speed: 69.663339 samples/sec	 accuracy=67.440000	 loss=1.323448	 lr=0.001000
Epoch[067] Batch [2549]/[3760]	Speed: 68.218663 samples/sec	 accuracy=67.462623	 loss=1.323026	 lr=0.001000
Epoch[067] Batch [2599]/[3760]	Speed: 70.022617 samples/sec	 accuracy=67.458534	 loss=1.323412	 lr=0.001000
Epoch[067] Batch [2649]/[3760]	Speed: 70.600632 samples/sec	 accuracy=67.462854	 loss=1.323438	 lr=0.001000
Epoch[067] Batch [2699]/[3760]	Speed: 69.739666 samples/sec	 accuracy=67.468750	 loss=1.322850	 lr=0.001000
Epoch[067] Batch [2749]/[3760]	Speed: 68.872721 samples/sec	 accuracy=67.433523	 loss=1.324516	 lr=0.001000
Epoch[067] Batch [2799]/[3760]	Speed: 68.587124 samples/sec	 accuracy=67.446429	 loss=1.323876	 lr=0.001000
Epoch[067] Batch [2849]/[3760]	Speed: 70.672740 samples/sec	 accuracy=67.447368	 loss=1.323955	 lr=0.001000
Epoch[067] Batch [2899]/[3760]	Speed: 68.749624 samples/sec	 accuracy=67.420259	 loss=1.325113	 lr=0.001000
Epoch[067] Batch [2949]/[3760]	Speed: 69.030761 samples/sec	 accuracy=67.404131	 loss=1.325622	 lr=0.001000
Epoch[067] Batch [2999]/[3760]	Speed: 69.917536 samples/sec	 accuracy=67.415104	 loss=1.325489	 lr=0.001000
Epoch[067] Batch [3049]/[3760]	Speed: 68.977667 samples/sec	 accuracy=67.390369	 loss=1.326264	 lr=0.001000
Epoch[067] Batch [3099]/[3760]	Speed: 69.658916 samples/sec	 accuracy=67.355847	 loss=1.326780	 lr=0.001000
Epoch[067] Batch [3149]/[3760]	Speed: 69.445482 samples/sec	 accuracy=67.364087	 loss=1.326316	 lr=0.001000
Epoch[067] Batch [3199]/[3760]	Speed: 70.187316 samples/sec	 accuracy=67.362793	 loss=1.326511	 lr=0.001000
Epoch[067] Batch [3249]/[3760]	Speed: 70.385575 samples/sec	 accuracy=67.352404	 loss=1.326981	 lr=0.001000
Epoch[067] Batch [3299]/[3760]	Speed: 66.292809 samples/sec	 accuracy=67.331913	 loss=1.327368	 lr=0.001000
Epoch[067] Batch [3349]/[3760]	Speed: 71.478328 samples/sec	 accuracy=67.324160	 loss=1.327684	 lr=0.001000
Epoch[067] Batch [3399]/[3760]	Speed: 69.804926 samples/sec	 accuracy=67.307445	 loss=1.328228	 lr=0.001000
Epoch[067] Batch [3449]/[3760]	Speed: 66.527377 samples/sec	 accuracy=67.303442	 loss=1.327993	 lr=0.001000
Epoch[067] Batch [3499]/[3760]	Speed: 73.184543 samples/sec	 accuracy=67.306696	 loss=1.328045	 lr=0.001000
Epoch[067] Batch [3549]/[3760]	Speed: 69.880606 samples/sec	 accuracy=67.312940	 loss=1.327883	 lr=0.001000
Epoch[067] Batch [3599]/[3760]	Speed: 68.638296 samples/sec	 accuracy=67.302083	 loss=1.328132	 lr=0.001000
Epoch[067] Batch [3649]/[3760]	Speed: 67.828089 samples/sec	 accuracy=67.297517	 loss=1.328165	 lr=0.001000
Epoch[067] Batch [3699]/[3760]	Speed: 71.689989 samples/sec	 accuracy=67.301520	 loss=1.328476	 lr=0.001000
Epoch[067] Batch [3749]/[3760]	Speed: 77.602166 samples/sec	 accuracy=67.292083	 loss=1.329126	 lr=0.001000
Batch [0049]/[0303]: acc-top1=62.125000 acc-top5=84.125000
Batch [0099]/[0303]: acc-top1=62.296875 acc-top5=83.906250
Batch [0149]/[0303]: acc-top1=62.843750 acc-top5=84.260417
Batch [0199]/[0303]: acc-top1=63.273437 acc-top5=84.664062
Batch [0249]/[0303]: acc-top1=63.375000 acc-top5=84.781250
Batch [0299]/[0303]: acc-top1=63.270833 acc-top5=84.645833
[Epoch 067] training: accuracy=67.299701	 loss=1.329041
[Epoch 067] speed: 69 samples/sec	time cost: 3757.454172
[Epoch 067] validation: acc-top1=63.237417 acc-top5=84.632838 loss=1.590874
Epoch[068] Batch [0049]/[3760]	Speed: 48.188031 samples/sec	 accuracy=66.968750	 loss=1.335373	 lr=0.001000
Epoch[068] Batch [0099]/[3760]	Speed: 69.733138 samples/sec	 accuracy=67.265625	 loss=1.319478	 lr=0.001000
Epoch[068] Batch [0149]/[3760]	Speed: 69.570121 samples/sec	 accuracy=67.770833	 loss=1.303633	 lr=0.001000
Epoch[068] Batch [0199]/[3760]	Speed: 68.537191 samples/sec	 accuracy=67.914062	 loss=1.308065	 lr=0.001000
Epoch[068] Batch [0249]/[3760]	Speed: 70.616221 samples/sec	 accuracy=68.193750	 loss=1.299876	 lr=0.001000
Epoch[068] Batch [0299]/[3760]	Speed: 69.693508 samples/sec	 accuracy=68.296875	 loss=1.297167	 lr=0.001000
Epoch[068] Batch [0349]/[3760]	Speed: 69.249911 samples/sec	 accuracy=67.986607	 loss=1.304318	 lr=0.001000
Epoch[068] Batch [0399]/[3760]	Speed: 69.568303 samples/sec	 accuracy=67.851562	 loss=1.307327	 lr=0.001000
Epoch[068] Batch [0449]/[3760]	Speed: 69.377718 samples/sec	 accuracy=67.750000	 loss=1.310615	 lr=0.001000
Epoch[068] Batch [0499]/[3760]	Speed: 69.127751 samples/sec	 accuracy=67.653125	 loss=1.309757	 lr=0.001000
Epoch[068] Batch [0549]/[3760]	Speed: 70.060373 samples/sec	 accuracy=67.769886	 loss=1.306950	 lr=0.001000
Epoch[068] Batch [0599]/[3760]	Speed: 71.057680 samples/sec	 accuracy=67.705729	 loss=1.312315	 lr=0.001000
Epoch[068] Batch [0649]/[3760]	Speed: 68.871665 samples/sec	 accuracy=67.677885	 loss=1.314113	 lr=0.001000
Epoch[068] Batch [0699]/[3760]	Speed: 70.887722 samples/sec	 accuracy=67.660714	 loss=1.314658	 lr=0.001000
Epoch[068] Batch [0749]/[3760]	Speed: 68.280157 samples/sec	 accuracy=67.685417	 loss=1.315055	 lr=0.001000
Epoch[068] Batch [0799]/[3760]	Speed: 70.450398 samples/sec	 accuracy=67.697266	 loss=1.317233	 lr=0.001000
Epoch[068] Batch [0849]/[3760]	Speed: 70.985819 samples/sec	 accuracy=67.599265	 loss=1.319882	 lr=0.001000
Epoch[068] Batch [0899]/[3760]	Speed: 68.627662 samples/sec	 accuracy=67.652778	 loss=1.317871	 lr=0.001000
Epoch[068] Batch [0949]/[3760]	Speed: 70.564559 samples/sec	 accuracy=67.679276	 loss=1.318220	 lr=0.001000
Epoch[068] Batch [0999]/[3760]	Speed: 70.398901 samples/sec	 accuracy=67.645313	 loss=1.318820	 lr=0.001000
Epoch[068] Batch [1049]/[3760]	Speed: 67.011955 samples/sec	 accuracy=67.724702	 loss=1.315532	 lr=0.001000
Epoch[068] Batch [1099]/[3760]	Speed: 71.900248 samples/sec	 accuracy=67.751420	 loss=1.315123	 lr=0.001000
Epoch[068] Batch [1149]/[3760]	Speed: 69.565566 samples/sec	 accuracy=67.767663	 loss=1.315093	 lr=0.001000
Epoch[068] Batch [1199]/[3760]	Speed: 68.527033 samples/sec	 accuracy=67.730469	 loss=1.315750	 lr=0.001000
Epoch[068] Batch [1249]/[3760]	Speed: 70.467644 samples/sec	 accuracy=67.692500	 loss=1.317153	 lr=0.001000
Epoch[068] Batch [1299]/[3760]	Speed: 70.795228 samples/sec	 accuracy=67.659856	 loss=1.317670	 lr=0.001000
Epoch[068] Batch [1349]/[3760]	Speed: 67.245986 samples/sec	 accuracy=67.659722	 loss=1.317111	 lr=0.001000
Epoch[068] Batch [1399]/[3760]	Speed: 71.164948 samples/sec	 accuracy=67.698661	 loss=1.316303	 lr=0.001000
Epoch[068] Batch [1449]/[3760]	Speed: 68.359446 samples/sec	 accuracy=67.728448	 loss=1.316090	 lr=0.001000
Epoch[068] Batch [1499]/[3760]	Speed: 71.139383 samples/sec	 accuracy=67.730208	 loss=1.315664	 lr=0.001000
Epoch[068] Batch [1549]/[3760]	Speed: 69.320921 samples/sec	 accuracy=67.708669	 loss=1.317132	 lr=0.001000
Epoch[068] Batch [1599]/[3760]	Speed: 70.050316 samples/sec	 accuracy=67.666992	 loss=1.318562	 lr=0.001000
Epoch[068] Batch [1649]/[3760]	Speed: 70.040188 samples/sec	 accuracy=67.629735	 loss=1.319625	 lr=0.001000
Epoch[068] Batch [1699]/[3760]	Speed: 69.693505 samples/sec	 accuracy=67.615809	 loss=1.319973	 lr=0.001000
Epoch[068] Batch [1749]/[3760]	Speed: 69.213204 samples/sec	 accuracy=67.594643	 loss=1.320992	 lr=0.001000
Epoch[068] Batch [1799]/[3760]	Speed: 68.971057 samples/sec	 accuracy=67.578125	 loss=1.322043	 lr=0.001000
Epoch[068] Batch [1849]/[3760]	Speed: 70.289462 samples/sec	 accuracy=67.548986	 loss=1.322494	 lr=0.001000
Epoch[068] Batch [1899]/[3760]	Speed: 70.897733 samples/sec	 accuracy=67.553454	 loss=1.322040	 lr=0.001000
Epoch[068] Batch [1949]/[3760]	Speed: 67.769949 samples/sec	 accuracy=67.545673	 loss=1.322665	 lr=0.001000
Epoch[068] Batch [1999]/[3760]	Speed: 67.987965 samples/sec	 accuracy=67.537500	 loss=1.322597	 lr=0.001000
Epoch[068] Batch [2049]/[3760]	Speed: 72.261748 samples/sec	 accuracy=67.532012	 loss=1.321981	 lr=0.001000
Epoch[068] Batch [2099]/[3760]	Speed: 69.376471 samples/sec	 accuracy=67.535714	 loss=1.321761	 lr=0.001000
Epoch[068] Batch [2149]/[3760]	Speed: 69.796069 samples/sec	 accuracy=67.513081	 loss=1.321709	 lr=0.001000
Epoch[068] Batch [2199]/[3760]	Speed: 69.576394 samples/sec	 accuracy=67.518466	 loss=1.322451	 lr=0.001000
Epoch[068] Batch [2249]/[3760]	Speed: 68.664784 samples/sec	 accuracy=67.546528	 loss=1.322460	 lr=0.001000
Epoch[068] Batch [2299]/[3760]	Speed: 68.822248 samples/sec	 accuracy=67.532609	 loss=1.323080	 lr=0.001000
Epoch[068] Batch [2349]/[3760]	Speed: 70.667460 samples/sec	 accuracy=67.507314	 loss=1.323722	 lr=0.001000
Epoch[068] Batch [2399]/[3760]	Speed: 70.627072 samples/sec	 accuracy=67.487630	 loss=1.324702	 lr=0.001000
Epoch[068] Batch [2449]/[3760]	Speed: 68.930561 samples/sec	 accuracy=67.514668	 loss=1.323675	 lr=0.001000
Epoch[068] Batch [2499]/[3760]	Speed: 69.189824 samples/sec	 accuracy=67.496250	 loss=1.324008	 lr=0.001000
Epoch[068] Batch [2549]/[3760]	Speed: 69.469010 samples/sec	 accuracy=67.503676	 loss=1.323752	 lr=0.001000
Epoch[068] Batch [2599]/[3760]	Speed: 71.807426 samples/sec	 accuracy=67.517428	 loss=1.323610	 lr=0.001000
Epoch[068] Batch [2649]/[3760]	Speed: 68.486089 samples/sec	 accuracy=67.502948	 loss=1.324337	 lr=0.001000
Epoch[068] Batch [2699]/[3760]	Speed: 70.317657 samples/sec	 accuracy=67.485532	 loss=1.325195	 lr=0.001000
Epoch[068] Batch [2749]/[3760]	Speed: 69.151029 samples/sec	 accuracy=67.499432	 loss=1.324915	 lr=0.001000
Epoch[068] Batch [2799]/[3760]	Speed: 70.327097 samples/sec	 accuracy=67.530692	 loss=1.323831	 lr=0.001000
Epoch[068] Batch [2849]/[3760]	Speed: 69.557135 samples/sec	 accuracy=67.500548	 loss=1.324730	 lr=0.001000
Epoch[068] Batch [2899]/[3760]	Speed: 69.522741 samples/sec	 accuracy=67.493534	 loss=1.325292	 lr=0.001000
Epoch[068] Batch [2949]/[3760]	Speed: 68.592511 samples/sec	 accuracy=67.475106	 loss=1.325704	 lr=0.001000
Epoch[068] Batch [2999]/[3760]	Speed: 69.720674 samples/sec	 accuracy=67.483854	 loss=1.325816	 lr=0.001000
Epoch[068] Batch [3049]/[3760]	Speed: 69.915032 samples/sec	 accuracy=67.457992	 loss=1.326621	 lr=0.001000
Epoch[068] Batch [3099]/[3760]	Speed: 68.681310 samples/sec	 accuracy=67.456149	 loss=1.326848	 lr=0.001000
Epoch[068] Batch [3149]/[3760]	Speed: 70.121699 samples/sec	 accuracy=67.437996	 loss=1.327519	 lr=0.001000
Epoch[068] Batch [3199]/[3760]	Speed: 69.531874 samples/sec	 accuracy=67.447266	 loss=1.327673	 lr=0.001000
Epoch[068] Batch [3249]/[3760]	Speed: 69.284904 samples/sec	 accuracy=67.452885	 loss=1.327303	 lr=0.001000
Epoch[068] Batch [3299]/[3760]	Speed: 71.114408 samples/sec	 accuracy=67.460227	 loss=1.327202	 lr=0.001000
Epoch[068] Batch [3349]/[3760]	Speed: 69.566529 samples/sec	 accuracy=67.454757	 loss=1.327531	 lr=0.001000
Epoch[068] Batch [3399]/[3760]	Speed: 69.105985 samples/sec	 accuracy=67.458640	 loss=1.327613	 lr=0.001000
Epoch[068] Batch [3449]/[3760]	Speed: 69.133878 samples/sec	 accuracy=67.463768	 loss=1.327124	 lr=0.001000
Epoch[068] Batch [3499]/[3760]	Speed: 70.123067 samples/sec	 accuracy=67.450446	 loss=1.327217	 lr=0.001000
Epoch[068] Batch [3549]/[3760]	Speed: 69.639918 samples/sec	 accuracy=67.437060	 loss=1.327646	 lr=0.001000
Epoch[068] Batch [3599]/[3760]	Speed: 70.031143 samples/sec	 accuracy=67.414931	 loss=1.328324	 lr=0.001000
Epoch[068] Batch [3649]/[3760]	Speed: 70.329853 samples/sec	 accuracy=67.413099	 loss=1.328633	 lr=0.001000
Epoch[068] Batch [3699]/[3760]	Speed: 68.862101 samples/sec	 accuracy=67.416807	 loss=1.328890	 lr=0.001000
Epoch[068] Batch [3749]/[3760]	Speed: 76.894968 samples/sec	 accuracy=67.411667	 loss=1.329374	 lr=0.001000
Batch [0049]/[0303]: acc-top1=61.812500 acc-top5=83.718750
Batch [0099]/[0303]: acc-top1=62.062500 acc-top5=83.562500
Batch [0149]/[0303]: acc-top1=62.864583 acc-top5=83.833333
Batch [0199]/[0303]: acc-top1=63.367188 acc-top5=84.343750
Batch [0249]/[0303]: acc-top1=63.587500 acc-top5=84.581250
Batch [0299]/[0303]: acc-top1=63.463542 acc-top5=84.473958
[Epoch 068] training: accuracy=67.418135	 loss=1.328946
[Epoch 068] speed: 69 samples/sec	time cost: 3753.957462
[Epoch 068] validation: acc-top1=63.454002 acc-top5=84.478135 loss=1.591362
Epoch[069] Batch [0049]/[3759]	Speed: 48.061303 samples/sec	 accuracy=67.187500	 loss=1.347607	 lr=0.001000
Epoch[069] Batch [0099]/[3759]	Speed: 70.047538 samples/sec	 accuracy=67.500000	 loss=1.326986	 lr=0.001000
Epoch[069] Batch [0149]/[3759]	Speed: 69.823160 samples/sec	 accuracy=67.531250	 loss=1.318867	 lr=0.001000
Epoch[069] Batch [0199]/[3759]	Speed: 70.068311 samples/sec	 accuracy=67.515625	 loss=1.316442	 lr=0.001000
Epoch[069] Batch [0249]/[3759]	Speed: 69.525662 samples/sec	 accuracy=67.618750	 loss=1.319108	 lr=0.001000
Epoch[069] Batch [0299]/[3759]	Speed: 68.568516 samples/sec	 accuracy=67.776042	 loss=1.313804	 lr=0.001000
Epoch[069] Batch [0349]/[3759]	Speed: 69.092711 samples/sec	 accuracy=67.794643	 loss=1.312839	 lr=0.001000
Epoch[069] Batch [0399]/[3759]	Speed: 71.431196 samples/sec	 accuracy=67.765625	 loss=1.314366	 lr=0.001000
Epoch[069] Batch [0449]/[3759]	Speed: 69.711652 samples/sec	 accuracy=67.711806	 loss=1.318410	 lr=0.001000
Epoch[069] Batch [0499]/[3759]	Speed: 70.681930 samples/sec	 accuracy=67.731250	 loss=1.320753	 lr=0.001000
Epoch[069] Batch [0549]/[3759]	Speed: 67.982294 samples/sec	 accuracy=67.840909	 loss=1.315892	 lr=0.001000
Epoch[069] Batch [0599]/[3759]	Speed: 70.737739 samples/sec	 accuracy=67.877604	 loss=1.314015	 lr=0.001000
Epoch[069] Batch [0649]/[3759]	Speed: 70.844948 samples/sec	 accuracy=67.858173	 loss=1.315453	 lr=0.001000
Epoch[069] Batch [0699]/[3759]	Speed: 69.094543 samples/sec	 accuracy=67.859375	 loss=1.316484	 lr=0.001000
Epoch[069] Batch [0749]/[3759]	Speed: 70.302559 samples/sec	 accuracy=67.829167	 loss=1.315337	 lr=0.001000
Epoch[069] Batch [0799]/[3759]	Speed: 68.302670 samples/sec	 accuracy=67.765625	 loss=1.315804	 lr=0.001000
Epoch[069] Batch [0849]/[3759]	Speed: 69.920923 samples/sec	 accuracy=67.707721	 loss=1.318384	 lr=0.001000
Epoch[069] Batch [0899]/[3759]	Speed: 69.488423 samples/sec	 accuracy=67.730903	 loss=1.317202	 lr=0.001000
Epoch[069] Batch [0949]/[3759]	Speed: 70.212783 samples/sec	 accuracy=67.771382	 loss=1.315597	 lr=0.001000
Epoch[069] Batch [0999]/[3759]	Speed: 69.658762 samples/sec	 accuracy=67.685938	 loss=1.319367	 lr=0.001000
Epoch[069] Batch [1049]/[3759]	Speed: 69.421201 samples/sec	 accuracy=67.691964	 loss=1.319046	 lr=0.001000
Epoch[069] Batch [1099]/[3759]	Speed: 70.598811 samples/sec	 accuracy=67.696023	 loss=1.319812	 lr=0.001000
Epoch[069] Batch [1149]/[3759]	Speed: 68.967517 samples/sec	 accuracy=67.729620	 loss=1.317537	 lr=0.001000
Epoch[069] Batch [1199]/[3759]	Speed: 66.897079 samples/sec	 accuracy=67.697917	 loss=1.318542	 lr=0.001000
Epoch[069] Batch [1249]/[3759]	Speed: 71.318083 samples/sec	 accuracy=67.648750	 loss=1.319820	 lr=0.001000
Epoch[069] Batch [1299]/[3759]	Speed: 70.698165 samples/sec	 accuracy=67.621394	 loss=1.319112	 lr=0.001000
Epoch[069] Batch [1349]/[3759]	Speed: 70.289387 samples/sec	 accuracy=67.630787	 loss=1.318612	 lr=0.001000
Epoch[069] Batch [1399]/[3759]	Speed: 68.958606 samples/sec	 accuracy=67.639509	 loss=1.318177	 lr=0.001000
Epoch[069] Batch [1449]/[3759]	Speed: 69.990458 samples/sec	 accuracy=67.630388	 loss=1.319544	 lr=0.001000
Epoch[069] Batch [1499]/[3759]	Speed: 68.407187 samples/sec	 accuracy=67.650000	 loss=1.318875	 lr=0.001000
Epoch[069] Batch [1549]/[3759]	Speed: 70.281091 samples/sec	 accuracy=67.605847	 loss=1.319701	 lr=0.001000
Epoch[069] Batch [1599]/[3759]	Speed: 69.751725 samples/sec	 accuracy=67.609375	 loss=1.319278	 lr=0.001000
Epoch[069] Batch [1649]/[3759]	Speed: 70.610286 samples/sec	 accuracy=67.615530	 loss=1.318945	 lr=0.001000
Epoch[069] Batch [1699]/[3759]	Speed: 70.288010 samples/sec	 accuracy=67.633272	 loss=1.317969	 lr=0.001000
Epoch[069] Batch [1749]/[3759]	Speed: 69.187337 samples/sec	 accuracy=67.631250	 loss=1.317241	 lr=0.001000
Epoch[069] Batch [1799]/[3759]	Speed: 69.658424 samples/sec	 accuracy=67.644965	 loss=1.316191	 lr=0.001000
Epoch[069] Batch [1849]/[3759]	Speed: 70.112574 samples/sec	 accuracy=67.628378	 loss=1.315426	 lr=0.001000
Epoch[069] Batch [1899]/[3759]	Speed: 70.330214 samples/sec	 accuracy=67.612664	 loss=1.315970	 lr=0.001000
Epoch[069] Batch [1949]/[3759]	Speed: 68.765695 samples/sec	 accuracy=67.641827	 loss=1.314863	 lr=0.001000
Epoch[069] Batch [1999]/[3759]	Speed: 70.689398 samples/sec	 accuracy=67.633594	 loss=1.314398	 lr=0.001000
Epoch[069] Batch [2049]/[3759]	Speed: 68.829101 samples/sec	 accuracy=67.635671	 loss=1.314265	 lr=0.001000
Epoch[069] Batch [2099]/[3759]	Speed: 69.638280 samples/sec	 accuracy=67.627232	 loss=1.314406	 lr=0.001000
Epoch[069] Batch [2149]/[3759]	Speed: 70.608559 samples/sec	 accuracy=67.611919	 loss=1.315148	 lr=0.001000
Epoch[069] Batch [2199]/[3759]	Speed: 69.498358 samples/sec	 accuracy=67.617898	 loss=1.315069	 lr=0.001000
Epoch[069] Batch [2249]/[3759]	Speed: 69.296834 samples/sec	 accuracy=67.612500	 loss=1.315254	 lr=0.001000
Epoch[069] Batch [2299]/[3759]	Speed: 68.714642 samples/sec	 accuracy=67.614130	 loss=1.315370	 lr=0.001000
Epoch[069] Batch [2349]/[3759]	Speed: 70.292723 samples/sec	 accuracy=67.571809	 loss=1.317105	 lr=0.001000
Epoch[069] Batch [2399]/[3759]	Speed: 70.436876 samples/sec	 accuracy=67.557943	 loss=1.317836	 lr=0.001000
Epoch[069] Batch [2449]/[3759]	Speed: 68.810952 samples/sec	 accuracy=67.566327	 loss=1.317001	 lr=0.001000
Epoch[069] Batch [2499]/[3759]	Speed: 71.200880 samples/sec	 accuracy=67.557500	 loss=1.317494	 lr=0.001000
Epoch[069] Batch [2549]/[3759]	Speed: 68.247727 samples/sec	 accuracy=67.526961	 loss=1.318885	 lr=0.001000
Epoch[069] Batch [2599]/[3759]	Speed: 70.561722 samples/sec	 accuracy=67.521635	 loss=1.318546	 lr=0.001000
Epoch[069] Batch [2649]/[3759]	Speed: 69.997891 samples/sec	 accuracy=67.524175	 loss=1.318471	 lr=0.001000
Epoch[069] Batch [2699]/[3759]	Speed: 70.041586 samples/sec	 accuracy=67.530093	 loss=1.318328	 lr=0.001000
Epoch[069] Batch [2749]/[3759]	Speed: 69.386295 samples/sec	 accuracy=67.530114	 loss=1.318953	 lr=0.001000
Epoch[069] Batch [2799]/[3759]	Speed: 69.904616 samples/sec	 accuracy=67.506696	 loss=1.319804	 lr=0.001000
Epoch[069] Batch [2849]/[3759]	Speed: 68.181043 samples/sec	 accuracy=67.508224	 loss=1.320176	 lr=0.001000
Epoch[069] Batch [2899]/[3759]	Speed: 70.960743 samples/sec	 accuracy=67.496228	 loss=1.321069	 lr=0.001000
Epoch[069] Batch [2949]/[3759]	Speed: 70.997602 samples/sec	 accuracy=67.490466	 loss=1.321620	 lr=0.001000
Epoch[069] Batch [2999]/[3759]	Speed: 68.505055 samples/sec	 accuracy=67.526042	 loss=1.320956	 lr=0.001000
Epoch[069] Batch [3049]/[3759]	Speed: 71.212709 samples/sec	 accuracy=67.506148	 loss=1.322131	 lr=0.001000
Epoch[069] Batch [3099]/[3759]	Speed: 67.508758 samples/sec	 accuracy=67.506048	 loss=1.322813	 lr=0.001000
Epoch[069] Batch [3149]/[3759]	Speed: 71.470226 samples/sec	 accuracy=67.500496	 loss=1.322963	 lr=0.001000
Epoch[069] Batch [3199]/[3759]	Speed: 69.123140 samples/sec	 accuracy=67.502930	 loss=1.323155	 lr=0.001000
Epoch[069] Batch [3249]/[3759]	Speed: 70.443374 samples/sec	 accuracy=67.507212	 loss=1.323244	 lr=0.001000
Epoch[069] Batch [3299]/[3759]	Speed: 69.760155 samples/sec	 accuracy=67.506629	 loss=1.323042	 lr=0.001000
Epoch[069] Batch [3349]/[3759]	Speed: 70.254876 samples/sec	 accuracy=67.489272	 loss=1.323494	 lr=0.001000
Epoch[069] Batch [3399]/[3759]	Speed: 70.277416 samples/sec	 accuracy=67.475643	 loss=1.324056	 lr=0.001000
Epoch[069] Batch [3449]/[3759]	Speed: 69.154782 samples/sec	 accuracy=67.460145	 loss=1.324441	 lr=0.001000
Epoch[069] Batch [3499]/[3759]	Speed: 70.717042 samples/sec	 accuracy=67.460268	 loss=1.324694	 lr=0.001000
Epoch[069] Batch [3549]/[3759]	Speed: 68.998435 samples/sec	 accuracy=67.457306	 loss=1.324603	 lr=0.001000
Epoch[069] Batch [3599]/[3759]	Speed: 69.298133 samples/sec	 accuracy=67.474392	 loss=1.324250	 lr=0.001000
Epoch[069] Batch [3649]/[3759]	Speed: 70.537016 samples/sec	 accuracy=67.482021	 loss=1.324170	 lr=0.001000
Epoch[069] Batch [3699]/[3759]	Speed: 70.610031 samples/sec	 accuracy=67.472128	 loss=1.324196	 lr=0.001000
Epoch[069] Batch [3749]/[3759]	Speed: 78.317177 samples/sec	 accuracy=67.482917	 loss=1.323257	 lr=0.001000
Batch [0049]/[0303]: acc-top1=61.562500 acc-top5=84.062500
Batch [0099]/[0303]: acc-top1=62.281250 acc-top5=83.953125
Batch [0149]/[0303]: acc-top1=63.072917 acc-top5=84.270833
Batch [0199]/[0303]: acc-top1=63.460938 acc-top5=84.679688
Batch [0249]/[0303]: acc-top1=63.537500 acc-top5=84.818750
Batch [0299]/[0303]: acc-top1=63.375000 acc-top5=84.614583
[Epoch 069] training: accuracy=67.478053	 loss=1.323437
[Epoch 069] speed: 69 samples/sec	time cost: 3747.660254
[Epoch 069] validation: acc-top1=63.366337 acc-top5=84.622525 loss=1.597813
Epoch[070] Batch [0049]/[3760]	Speed: 48.005852 samples/sec	 accuracy=68.281250	 loss=1.301850	 lr=0.001000
Epoch[070] Batch [0099]/[3760]	Speed: 67.525865 samples/sec	 accuracy=67.937500	 loss=1.320866	 lr=0.001000
Epoch[070] Batch [0149]/[3760]	Speed: 69.926507 samples/sec	 accuracy=68.083333	 loss=1.299006	 lr=0.001000
Epoch[070] Batch [0199]/[3760]	Speed: 69.497552 samples/sec	 accuracy=67.992188	 loss=1.296795	 lr=0.001000
Epoch[070] Batch [0249]/[3760]	Speed: 70.789972 samples/sec	 accuracy=68.081250	 loss=1.293506	 lr=0.001000
Epoch[070] Batch [0299]/[3760]	Speed: 69.388192 samples/sec	 accuracy=67.942708	 loss=1.303219	 lr=0.001000
Epoch[070] Batch [0349]/[3760]	Speed: 69.191435 samples/sec	 accuracy=67.848214	 loss=1.308883	 lr=0.001000
Epoch[070] Batch [0399]/[3760]	Speed: 69.954467 samples/sec	 accuracy=67.792969	 loss=1.310949	 lr=0.001000
Epoch[070] Batch [0449]/[3760]	Speed: 68.749205 samples/sec	 accuracy=67.968750	 loss=1.308490	 lr=0.001000
Epoch[070] Batch [0499]/[3760]	Speed: 70.492833 samples/sec	 accuracy=68.031250	 loss=1.308430	 lr=0.001000
Epoch[070] Batch [0549]/[3760]	Speed: 68.825467 samples/sec	 accuracy=67.968750	 loss=1.310882	 lr=0.001000
Epoch[070] Batch [0599]/[3760]	Speed: 69.905511 samples/sec	 accuracy=67.950521	 loss=1.310774	 lr=0.001000
Epoch[070] Batch [0649]/[3760]	Speed: 70.915509 samples/sec	 accuracy=67.887019	 loss=1.314953	 lr=0.001000
Epoch[070] Batch [0699]/[3760]	Speed: 69.713592 samples/sec	 accuracy=67.857143	 loss=1.314145	 lr=0.001000
Epoch[070] Batch [0749]/[3760]	Speed: 68.667165 samples/sec	 accuracy=67.741667	 loss=1.319572	 lr=0.001000
Epoch[070] Batch [0799]/[3760]	Speed: 69.929657 samples/sec	 accuracy=67.787109	 loss=1.317325	 lr=0.001000
Epoch[070] Batch [0849]/[3760]	Speed: 69.428415 samples/sec	 accuracy=67.816176	 loss=1.316332	 lr=0.001000
Epoch[070] Batch [0899]/[3760]	Speed: 70.145080 samples/sec	 accuracy=67.819444	 loss=1.316296	 lr=0.001000
Epoch[070] Batch [0949]/[3760]	Speed: 69.218995 samples/sec	 accuracy=67.799342	 loss=1.316806	 lr=0.001000
Epoch[070] Batch [0999]/[3760]	Speed: 70.513171 samples/sec	 accuracy=67.843750	 loss=1.313878	 lr=0.001000
Epoch[070] Batch [1049]/[3760]	Speed: 69.891626 samples/sec	 accuracy=67.834821	 loss=1.315563	 lr=0.001000
Epoch[070] Batch [1099]/[3760]	Speed: 70.332546 samples/sec	 accuracy=67.823864	 loss=1.316464	 lr=0.001000
Epoch[070] Batch [1149]/[3760]	Speed: 67.677387 samples/sec	 accuracy=67.864130	 loss=1.316045	 lr=0.001000
Epoch[070] Batch [1199]/[3760]	Speed: 71.452483 samples/sec	 accuracy=67.871094	 loss=1.315571	 lr=0.001000
Epoch[070] Batch [1249]/[3760]	Speed: 68.381133 samples/sec	 accuracy=67.887500	 loss=1.314244	 lr=0.001000
Epoch[070] Batch [1299]/[3760]	Speed: 69.987509 samples/sec	 accuracy=67.877404	 loss=1.314548	 lr=0.001000
Epoch[070] Batch [1349]/[3760]	Speed: 70.195606 samples/sec	 accuracy=67.849537	 loss=1.315489	 lr=0.001000
Epoch[070] Batch [1399]/[3760]	Speed: 70.259164 samples/sec	 accuracy=67.891741	 loss=1.313060	 lr=0.001000
Epoch[070] Batch [1449]/[3760]	Speed: 68.488396 samples/sec	 accuracy=67.926724	 loss=1.311285	 lr=0.001000
Epoch[070] Batch [1499]/[3760]	Speed: 72.177981 samples/sec	 accuracy=67.901042	 loss=1.311765	 lr=0.001000
Epoch[070] Batch [1549]/[3760]	Speed: 69.977854 samples/sec	 accuracy=67.876008	 loss=1.312457	 lr=0.001000
Epoch[070] Batch [1599]/[3760]	Speed: 69.241768 samples/sec	 accuracy=67.907227	 loss=1.311622	 lr=0.001000
Epoch[070] Batch [1649]/[3760]	Speed: 69.976108 samples/sec	 accuracy=67.878788	 loss=1.310915	 lr=0.001000
Epoch[070] Batch [1699]/[3760]	Speed: 69.126941 samples/sec	 accuracy=67.863971	 loss=1.310533	 lr=0.001000
Epoch[070] Batch [1749]/[3760]	Speed: 71.004998 samples/sec	 accuracy=67.895536	 loss=1.309394	 lr=0.001000
Epoch[070] Batch [1799]/[3760]	Speed: 68.583187 samples/sec	 accuracy=67.898438	 loss=1.309578	 lr=0.001000
Epoch[070] Batch [1849]/[3760]	Speed: 70.930108 samples/sec	 accuracy=67.884291	 loss=1.309317	 lr=0.001000
Epoch[070] Batch [1899]/[3760]	Speed: 71.172435 samples/sec	 accuracy=67.870066	 loss=1.309443	 lr=0.001000
Epoch[070] Batch [1949]/[3760]	Speed: 69.291167 samples/sec	 accuracy=67.858173	 loss=1.310324	 lr=0.001000
Epoch[070] Batch [1999]/[3760]	Speed: 69.895058 samples/sec	 accuracy=67.817969	 loss=1.311905	 lr=0.001000
Epoch[070] Batch [2049]/[3760]	Speed: 70.179272 samples/sec	 accuracy=67.814024	 loss=1.311805	 lr=0.001000
Epoch[070] Batch [2099]/[3760]	Speed: 69.854770 samples/sec	 accuracy=67.787946	 loss=1.312469	 lr=0.001000
Epoch[070] Batch [2149]/[3760]	Speed: 68.531593 samples/sec	 accuracy=67.771802	 loss=1.312853	 lr=0.001000
Epoch[070] Batch [2199]/[3760]	Speed: 69.078426 samples/sec	 accuracy=67.750710	 loss=1.312949	 lr=0.001000
Epoch[070] Batch [2249]/[3760]	Speed: 71.059283 samples/sec	 accuracy=67.757639	 loss=1.312427	 lr=0.001000
Epoch[070] Batch [2299]/[3760]	Speed: 69.853511 samples/sec	 accuracy=67.747962	 loss=1.313970	 lr=0.001000
Epoch[070] Batch [2349]/[3760]	Speed: 68.906885 samples/sec	 accuracy=67.743351	 loss=1.314300	 lr=0.001000
Epoch[070] Batch [2399]/[3760]	Speed: 71.000742 samples/sec	 accuracy=67.720052	 loss=1.314805	 lr=0.001000
Epoch[070] Batch [2449]/[3760]	Speed: 68.925067 samples/sec	 accuracy=67.715561	 loss=1.315313	 lr=0.001000
Epoch[070] Batch [2499]/[3760]	Speed: 67.542888 samples/sec	 accuracy=67.716250	 loss=1.315891	 lr=0.001000
Epoch[070] Batch [2549]/[3760]	Speed: 68.092054 samples/sec	 accuracy=67.712623	 loss=1.316249	 lr=0.001000
Epoch[070] Batch [2599]/[3760]	Speed: 72.883750 samples/sec	 accuracy=67.713942	 loss=1.316137	 lr=0.001000
Epoch[070] Batch [2649]/[3760]	Speed: 69.717592 samples/sec	 accuracy=67.709906	 loss=1.315830	 lr=0.001000
Epoch[070] Batch [2699]/[3760]	Speed: 67.466682 samples/sec	 accuracy=67.704861	 loss=1.315331	 lr=0.001000
Epoch[070] Batch [2749]/[3760]	Speed: 71.328424 samples/sec	 accuracy=67.703409	 loss=1.315292	 lr=0.001000
Epoch[070] Batch [2799]/[3760]	Speed: 70.304682 samples/sec	 accuracy=67.683594	 loss=1.315962	 lr=0.001000
Epoch[070] Batch [2849]/[3760]	Speed: 68.837535 samples/sec	 accuracy=67.665022	 loss=1.316384	 lr=0.001000
Epoch[070] Batch [2899]/[3760]	Speed: 70.618571 samples/sec	 accuracy=67.674569	 loss=1.316212	 lr=0.001000
Epoch[070] Batch [2949]/[3760]	Speed: 71.484339 samples/sec	 accuracy=67.685381	 loss=1.315553	 lr=0.001000
Epoch[070] Batch [2999]/[3760]	Speed: 68.994528 samples/sec	 accuracy=67.706771	 loss=1.314685	 lr=0.001000
Epoch[070] Batch [3049]/[3760]	Speed: 68.736105 samples/sec	 accuracy=67.709529	 loss=1.314515	 lr=0.001000
Epoch[070] Batch [3099]/[3760]	Speed: 70.091960 samples/sec	 accuracy=67.694052	 loss=1.314939	 lr=0.001000
Epoch[070] Batch [3149]/[3760]	Speed: 68.394270 samples/sec	 accuracy=67.693948	 loss=1.314970	 lr=0.001000
Epoch[070] Batch [3199]/[3760]	Speed: 70.060994 samples/sec	 accuracy=67.666992	 loss=1.315928	 lr=0.001000
Epoch[070] Batch [3249]/[3760]	Speed: 69.910311 samples/sec	 accuracy=67.662019	 loss=1.316139	 lr=0.001000
Epoch[070] Batch [3299]/[3760]	Speed: 69.851514 samples/sec	 accuracy=67.649621	 loss=1.316527	 lr=0.001000
Epoch[070] Batch [3349]/[3760]	Speed: 69.811212 samples/sec	 accuracy=67.649254	 loss=1.316339	 lr=0.001000
Epoch[070] Batch [3399]/[3760]	Speed: 70.423778 samples/sec	 accuracy=67.642463	 loss=1.316489	 lr=0.001000
Epoch[070] Batch [3449]/[3760]	Speed: 70.267043 samples/sec	 accuracy=67.627264	 loss=1.316962	 lr=0.001000
Epoch[070] Batch [3499]/[3760]	Speed: 68.325348 samples/sec	 accuracy=67.628125	 loss=1.316357	 lr=0.001000
Epoch[070] Batch [3549]/[3760]	Speed: 71.621775 samples/sec	 accuracy=67.615317	 loss=1.316777	 lr=0.001000
Epoch[070] Batch [3599]/[3760]	Speed: 68.640965 samples/sec	 accuracy=67.622396	 loss=1.316919	 lr=0.001000
Epoch[070] Batch [3649]/[3760]	Speed: 67.372347 samples/sec	 accuracy=67.632705	 loss=1.316784	 lr=0.001000
Epoch[070] Batch [3699]/[3760]	Speed: 70.788873 samples/sec	 accuracy=67.631334	 loss=1.316589	 lr=0.001000
Epoch[070] Batch [3749]/[3760]	Speed: 78.642843 samples/sec	 accuracy=67.632917	 loss=1.316651	 lr=0.001000
Batch [0049]/[0303]: acc-top1=61.375000 acc-top5=83.343750
Batch [0099]/[0303]: acc-top1=61.687500 acc-top5=83.453125
Batch [0149]/[0303]: acc-top1=62.656250 acc-top5=84.104167
Batch [0199]/[0303]: acc-top1=63.093750 acc-top5=84.562500
Batch [0249]/[0303]: acc-top1=63.350000 acc-top5=84.718750
Batch [0299]/[0303]: acc-top1=63.130208 acc-top5=84.546875
[Epoch 070] training: accuracy=67.637550	 loss=1.316441
[Epoch 070] speed: 69 samples/sec	time cost: 3751.204193
[Epoch 070] validation: acc-top1=63.108498 acc-top5=84.550330 loss=1.604785
Epoch[071] Batch [0049]/[3760]	Speed: 47.750890 samples/sec	 accuracy=66.687500	 loss=1.327222	 lr=0.001000
Epoch[071] Batch [0099]/[3760]	Speed: 68.944062 samples/sec	 accuracy=67.062500	 loss=1.323568	 lr=0.001000
Epoch[071] Batch [0149]/[3760]	Speed: 69.302717 samples/sec	 accuracy=67.177083	 loss=1.313638	 lr=0.001000
Epoch[071] Batch [0199]/[3760]	Speed: 69.312991 samples/sec	 accuracy=67.117188	 loss=1.329805	 lr=0.001000
Epoch[071] Batch [0249]/[3760]	Speed: 70.171966 samples/sec	 accuracy=67.225000	 loss=1.326185	 lr=0.001000
Epoch[071] Batch [0299]/[3760]	Speed: 70.671534 samples/sec	 accuracy=67.442708	 loss=1.314370	 lr=0.001000
Epoch[071] Batch [0349]/[3760]	Speed: 68.164529 samples/sec	 accuracy=67.616071	 loss=1.312651	 lr=0.001000
Epoch[071] Batch [0399]/[3760]	Speed: 69.730382 samples/sec	 accuracy=67.597656	 loss=1.313623	 lr=0.001000
Epoch[071] Batch [0449]/[3760]	Speed: 71.202565 samples/sec	 accuracy=67.687500	 loss=1.309316	 lr=0.001000
Epoch[071] Batch [0499]/[3760]	Speed: 68.881084 samples/sec	 accuracy=67.581250	 loss=1.312038	 lr=0.001000
Epoch[071] Batch [0549]/[3760]	Speed: 70.800609 samples/sec	 accuracy=67.542614	 loss=1.315264	 lr=0.001000
Epoch[071] Batch [0599]/[3760]	Speed: 68.121890 samples/sec	 accuracy=67.554688	 loss=1.315485	 lr=0.001000
Epoch[071] Batch [0649]/[3760]	Speed: 71.366514 samples/sec	 accuracy=67.550481	 loss=1.315036	 lr=0.001000
Epoch[071] Batch [0699]/[3760]	Speed: 71.265454 samples/sec	 accuracy=67.544643	 loss=1.315902	 lr=0.001000
Epoch[071] Batch [0749]/[3760]	Speed: 67.469485 samples/sec	 accuracy=67.650000	 loss=1.310426	 lr=0.001000
Epoch[071] Batch [0799]/[3760]	Speed: 71.111682 samples/sec	 accuracy=67.648438	 loss=1.309310	 lr=0.001000
Epoch[071] Batch [0849]/[3760]	Speed: 69.167804 samples/sec	 accuracy=67.744485	 loss=1.307159	 lr=0.001000
Epoch[071] Batch [0899]/[3760]	Speed: 70.790839 samples/sec	 accuracy=67.756944	 loss=1.307930	 lr=0.001000
Epoch[071] Batch [0949]/[3760]	Speed: 69.155796 samples/sec	 accuracy=67.796053	 loss=1.308871	 lr=0.001000
Epoch[071] Batch [0999]/[3760]	Speed: 68.138854 samples/sec	 accuracy=67.760937	 loss=1.309703	 lr=0.001000
Epoch[071] Batch [1049]/[3760]	Speed: 70.103895 samples/sec	 accuracy=67.778274	 loss=1.308390	 lr=0.001000
Epoch[071] Batch [1099]/[3760]	Speed: 70.516887 samples/sec	 accuracy=67.754261	 loss=1.309811	 lr=0.001000
Epoch[071] Batch [1149]/[3760]	Speed: 68.008479 samples/sec	 accuracy=67.842391	 loss=1.308814	 lr=0.001000
Epoch[071] Batch [1199]/[3760]	Speed: 69.410429 samples/sec	 accuracy=67.825521	 loss=1.309923	 lr=0.001000
Epoch[071] Batch [1249]/[3760]	Speed: 69.359937 samples/sec	 accuracy=67.836250	 loss=1.309600	 lr=0.001000
Epoch[071] Batch [1299]/[3760]	Speed: 70.642423 samples/sec	 accuracy=67.778846	 loss=1.312290	 lr=0.001000
Epoch[071] Batch [1349]/[3760]	Speed: 68.108913 samples/sec	 accuracy=67.799769	 loss=1.311825	 lr=0.001000
Epoch[071] Batch [1399]/[3760]	Speed: 70.967763 samples/sec	 accuracy=67.831473	 loss=1.310989	 lr=0.001000
Epoch[071] Batch [1449]/[3760]	Speed: 69.663234 samples/sec	 accuracy=67.790948	 loss=1.311059	 lr=0.001000
Epoch[071] Batch [1499]/[3760]	Speed: 69.729610 samples/sec	 accuracy=67.760417	 loss=1.311128	 lr=0.001000
Epoch[071] Batch [1549]/[3760]	Speed: 68.636378 samples/sec	 accuracy=67.738911	 loss=1.311805	 lr=0.001000
Epoch[071] Batch [1599]/[3760]	Speed: 69.986564 samples/sec	 accuracy=67.745117	 loss=1.310811	 lr=0.001000
Epoch[071] Batch [1649]/[3760]	Speed: 71.493575 samples/sec	 accuracy=67.743371	 loss=1.310358	 lr=0.001000
Epoch[071] Batch [1699]/[3760]	Speed: 69.633435 samples/sec	 accuracy=67.695772	 loss=1.311895	 lr=0.001000
Epoch[071] Batch [1749]/[3760]	Speed: 68.679478 samples/sec	 accuracy=67.715179	 loss=1.311278	 lr=0.001000
Epoch[071] Batch [1799]/[3760]	Speed: 69.700406 samples/sec	 accuracy=67.711806	 loss=1.311273	 lr=0.001000
Epoch[071] Batch [1849]/[3760]	Speed: 68.320999 samples/sec	 accuracy=67.708615	 loss=1.310414	 lr=0.001000
Epoch[071] Batch [1899]/[3760]	Speed: 69.806026 samples/sec	 accuracy=67.740132	 loss=1.309118	 lr=0.001000
Epoch[071] Batch [1949]/[3760]	Speed: 71.129051 samples/sec	 accuracy=67.755609	 loss=1.308836	 lr=0.001000
Epoch[071] Batch [1999]/[3760]	Speed: 68.750665 samples/sec	 accuracy=67.791406	 loss=1.307717	 lr=0.001000
Epoch[071] Batch [2049]/[3760]	Speed: 68.595457 samples/sec	 accuracy=67.762195	 loss=1.308795	 lr=0.001000
Epoch[071] Batch [2099]/[3760]	Speed: 68.034517 samples/sec	 accuracy=67.747024	 loss=1.309698	 lr=0.001000
Epoch[071] Batch [2149]/[3760]	Speed: 72.418537 samples/sec	 accuracy=67.719477	 loss=1.311482	 lr=0.001000
Epoch[071] Batch [2199]/[3760]	Speed: 69.598708 samples/sec	 accuracy=67.731534	 loss=1.312158	 lr=0.001000
Epoch[071] Batch [2249]/[3760]	Speed: 67.956935 samples/sec	 accuracy=67.679861	 loss=1.313649	 lr=0.001000
Epoch[071] Batch [2299]/[3760]	Speed: 71.009839 samples/sec	 accuracy=67.699728	 loss=1.312936	 lr=0.001000
Epoch[071] Batch [2349]/[3760]	Speed: 68.430447 samples/sec	 accuracy=67.674202	 loss=1.314169	 lr=0.001000
Epoch[071] Batch [2399]/[3760]	Speed: 70.711759 samples/sec	 accuracy=67.692057	 loss=1.313650	 lr=0.001000
Epoch[071] Batch [2449]/[3760]	Speed: 70.081495 samples/sec	 accuracy=67.705995	 loss=1.313164	 lr=0.001000
Epoch[071] Batch [2499]/[3760]	Speed: 69.598561 samples/sec	 accuracy=67.686250	 loss=1.313494	 lr=0.001000
Epoch[071] Batch [2549]/[3760]	Speed: 69.106675 samples/sec	 accuracy=67.663603	 loss=1.313840	 lr=0.001000
Epoch[071] Batch [2599]/[3760]	Speed: 70.670041 samples/sec	 accuracy=67.680889	 loss=1.313642	 lr=0.001000
Epoch[071] Batch [2649]/[3760]	Speed: 69.965254 samples/sec	 accuracy=67.682783	 loss=1.313413	 lr=0.001000
Epoch[071] Batch [2699]/[3760]	Speed: 69.746891 samples/sec	 accuracy=67.715278	 loss=1.312364	 lr=0.001000
Epoch[071] Batch [2749]/[3760]	Speed: 70.083645 samples/sec	 accuracy=67.734091	 loss=1.312078	 lr=0.001000
Epoch[071] Batch [2799]/[3760]	Speed: 68.384533 samples/sec	 accuracy=67.747210	 loss=1.311711	 lr=0.001000
Epoch[071] Batch [2849]/[3760]	Speed: 70.521056 samples/sec	 accuracy=67.737939	 loss=1.312486	 lr=0.001000
Epoch[071] Batch [2899]/[3760]	Speed: 70.022193 samples/sec	 accuracy=67.737608	 loss=1.312429	 lr=0.001000
Epoch[071] Batch [2949]/[3760]	Speed: 68.904477 samples/sec	 accuracy=67.740466	 loss=1.312341	 lr=0.001000
Epoch[071] Batch [2999]/[3760]	Speed: 68.133002 samples/sec	 accuracy=67.728125	 loss=1.313074	 lr=0.001000
Epoch[071] Batch [3049]/[3760]	Speed: 69.503956 samples/sec	 accuracy=67.732070	 loss=1.313027	 lr=0.001000
Epoch[071] Batch [3099]/[3760]	Speed: 70.068172 samples/sec	 accuracy=67.732863	 loss=1.312779	 lr=0.001000
Epoch[071] Batch [3149]/[3760]	Speed: 66.718240 samples/sec	 accuracy=67.749008	 loss=1.312211	 lr=0.001000
Epoch[071] Batch [3199]/[3760]	Speed: 69.053046 samples/sec	 accuracy=67.751465	 loss=1.312257	 lr=0.001000
Epoch[071] Batch [3249]/[3760]	Speed: 68.401048 samples/sec	 accuracy=67.734615	 loss=1.312313	 lr=0.001000
Epoch[071] Batch [3299]/[3760]	Speed: 73.121424 samples/sec	 accuracy=67.745739	 loss=1.312103	 lr=0.001000
Epoch[071] Batch [3349]/[3760]	Speed: 69.284709 samples/sec	 accuracy=67.740205	 loss=1.312014	 lr=0.001000
Epoch[071] Batch [3399]/[3760]	Speed: 70.604967 samples/sec	 accuracy=67.737592	 loss=1.311901	 lr=0.001000
Epoch[071] Batch [3449]/[3760]	Speed: 69.245243 samples/sec	 accuracy=67.723732	 loss=1.312330	 lr=0.001000
Epoch[071] Batch [3499]/[3760]	Speed: 69.045894 samples/sec	 accuracy=67.721875	 loss=1.312397	 lr=0.001000
Epoch[071] Batch [3549]/[3760]	Speed: 70.080878 samples/sec	 accuracy=67.708627	 loss=1.312639	 lr=0.001000
Epoch[071] Batch [3599]/[3760]	Speed: 68.638960 samples/sec	 accuracy=67.703559	 loss=1.312609	 lr=0.001000
Epoch[071] Batch [3649]/[3760]	Speed: 70.425279 samples/sec	 accuracy=67.702055	 loss=1.312573	 lr=0.001000
Epoch[071] Batch [3699]/[3760]	Speed: 70.972253 samples/sec	 accuracy=67.690456	 loss=1.312897	 lr=0.001000
Epoch[071] Batch [3749]/[3760]	Speed: 75.948179 samples/sec	 accuracy=67.702917	 loss=1.312892	 lr=0.001000
Batch [0049]/[0303]: acc-top1=61.250000 acc-top5=83.593750
Batch [0099]/[0303]: acc-top1=61.640625 acc-top5=83.421875
Batch [0149]/[0303]: acc-top1=62.385417 acc-top5=83.750000
Batch [0199]/[0303]: acc-top1=63.015625 acc-top5=84.187500
Batch [0249]/[0303]: acc-top1=63.318750 acc-top5=84.493750
Batch [0299]/[0303]: acc-top1=63.161458 acc-top5=84.369792
[Epoch 071] training: accuracy=67.709857	 loss=1.312624
[Epoch 071] speed: 69 samples/sec	time cost: 3757.031750
[Epoch 071] validation: acc-top1=63.154909 acc-top5=84.385314 loss=1.597955
Epoch[072] Batch [0049]/[3759]	Speed: 47.908618 samples/sec	 accuracy=67.187500	 loss=1.323206	 lr=0.001000
Epoch[072] Batch [0099]/[3759]	Speed: 69.854707 samples/sec	 accuracy=67.140625	 loss=1.326267	 lr=0.001000
Epoch[072] Batch [0149]/[3759]	Speed: 68.607822 samples/sec	 accuracy=67.031250	 loss=1.328886	 lr=0.001000
Epoch[072] Batch [0199]/[3759]	Speed: 69.664143 samples/sec	 accuracy=67.015625	 loss=1.342142	 lr=0.001000
Epoch[072] Batch [0249]/[3759]	Speed: 70.300264 samples/sec	 accuracy=67.375000	 loss=1.330356	 lr=0.001000
Epoch[072] Batch [0299]/[3759]	Speed: 69.269450 samples/sec	 accuracy=67.593750	 loss=1.321742	 lr=0.001000
Epoch[072] Batch [0349]/[3759]	Speed: 70.280006 samples/sec	 accuracy=67.843750	 loss=1.316356	 lr=0.001000
Epoch[072] Batch [0399]/[3759]	Speed: 69.684886 samples/sec	 accuracy=67.796875	 loss=1.316402	 lr=0.001000
Epoch[072] Batch [0449]/[3759]	Speed: 68.397909 samples/sec	 accuracy=67.795139	 loss=1.318262	 lr=0.001000
Epoch[072] Batch [0499]/[3759]	Speed: 69.743467 samples/sec	 accuracy=67.796875	 loss=1.318513	 lr=0.001000
Epoch[072] Batch [0549]/[3759]	Speed: 70.395180 samples/sec	 accuracy=67.875000	 loss=1.314215	 lr=0.001000
Epoch[072] Batch [0599]/[3759]	Speed: 69.904101 samples/sec	 accuracy=67.958333	 loss=1.308977	 lr=0.001000
Epoch[072] Batch [0649]/[3759]	Speed: 69.370644 samples/sec	 accuracy=67.915865	 loss=1.312319	 lr=0.001000
Epoch[072] Batch [0699]/[3759]	Speed: 68.611049 samples/sec	 accuracy=67.850446	 loss=1.315033	 lr=0.001000
Epoch[072] Batch [0749]/[3759]	Speed: 70.907444 samples/sec	 accuracy=67.760417	 loss=1.317723	 lr=0.001000
Epoch[072] Batch [0799]/[3759]	Speed: 70.757267 samples/sec	 accuracy=67.792969	 loss=1.317274	 lr=0.001000
Epoch[072] Batch [0849]/[3759]	Speed: 69.127111 samples/sec	 accuracy=67.757353	 loss=1.317815	 lr=0.001000
Epoch[072] Batch [0899]/[3759]	Speed: 70.833876 samples/sec	 accuracy=67.767361	 loss=1.317856	 lr=0.001000
Epoch[072] Batch [0949]/[3759]	Speed: 70.117749 samples/sec	 accuracy=67.794408	 loss=1.315970	 lr=0.001000
Epoch[072] Batch [0999]/[3759]	Speed: 67.944572 samples/sec	 accuracy=67.750000	 loss=1.317441	 lr=0.001000
Epoch[072] Batch [1049]/[3759]	Speed: 70.165955 samples/sec	 accuracy=67.729167	 loss=1.319119	 lr=0.001000
Epoch[072] Batch [1099]/[3759]	Speed: 70.703161 samples/sec	 accuracy=67.690341	 loss=1.320571	 lr=0.001000
Epoch[072] Batch [1149]/[3759]	Speed: 70.223143 samples/sec	 accuracy=67.586957	 loss=1.323666	 lr=0.001000
Epoch[072] Batch [1199]/[3759]	Speed: 69.504848 samples/sec	 accuracy=67.618490	 loss=1.323511	 lr=0.001000
Epoch[072] Batch [1249]/[3759]	Speed: 69.491014 samples/sec	 accuracy=67.631250	 loss=1.322964	 lr=0.001000
Epoch[072] Batch [1299]/[3759]	Speed: 70.742491 samples/sec	 accuracy=67.638221	 loss=1.323922	 lr=0.001000
Epoch[072] Batch [1349]/[3759]	Speed: 67.743884 samples/sec	 accuracy=67.599537	 loss=1.325541	 lr=0.001000
Epoch[072] Batch [1399]/[3759]	Speed: 71.064836 samples/sec	 accuracy=67.574777	 loss=1.326862	 lr=0.001000
Epoch[072] Batch [1449]/[3759]	Speed: 69.700701 samples/sec	 accuracy=67.575431	 loss=1.326274	 lr=0.001000
Epoch[072] Batch [1499]/[3759]	Speed: 69.911025 samples/sec	 accuracy=67.566667	 loss=1.324999	 lr=0.001000
Epoch[072] Batch [1549]/[3759]	Speed: 70.260454 samples/sec	 accuracy=67.608871	 loss=1.322198	 lr=0.001000
Epoch[072] Batch [1599]/[3759]	Speed: 70.036460 samples/sec	 accuracy=67.586914	 loss=1.322729	 lr=0.001000
Epoch[072] Batch [1649]/[3759]	Speed: 68.109058 samples/sec	 accuracy=67.607008	 loss=1.322158	 lr=0.001000
Epoch[072] Batch [1699]/[3759]	Speed: 71.274022 samples/sec	 accuracy=67.621324	 loss=1.322222	 lr=0.001000
Epoch[072] Batch [1749]/[3759]	Speed: 70.107249 samples/sec	 accuracy=67.608036	 loss=1.321788	 lr=0.001000
Epoch[072] Batch [1799]/[3759]	Speed: 70.126858 samples/sec	 accuracy=67.605035	 loss=1.321865	 lr=0.001000
Epoch[072] Batch [1849]/[3759]	Speed: 67.983092 samples/sec	 accuracy=67.635980	 loss=1.320696	 lr=0.001000
Epoch[072] Batch [1899]/[3759]	Speed: 71.076064 samples/sec	 accuracy=67.599507	 loss=1.322020	 lr=0.001000
Epoch[072] Batch [1949]/[3759]	Speed: 70.342556 samples/sec	 accuracy=67.616987	 loss=1.321505	 lr=0.001000
Epoch[072] Batch [1999]/[3759]	Speed: 69.131815 samples/sec	 accuracy=67.607031	 loss=1.321686	 lr=0.001000
Epoch[072] Batch [2049]/[3759]	Speed: 68.601808 samples/sec	 accuracy=67.591463	 loss=1.320876	 lr=0.001000
Epoch[072] Batch [2099]/[3759]	Speed: 70.515646 samples/sec	 accuracy=67.618304	 loss=1.320402	 lr=0.001000
Epoch[072] Batch [2149]/[3759]	Speed: 69.963714 samples/sec	 accuracy=67.620640	 loss=1.319981	 lr=0.001000
Epoch[072] Batch [2199]/[3759]	Speed: 70.525082 samples/sec	 accuracy=67.642045	 loss=1.318842	 lr=0.001000
Epoch[072] Batch [2249]/[3759]	Speed: 69.870339 samples/sec	 accuracy=67.631944	 loss=1.319104	 lr=0.001000
Epoch[072] Batch [2299]/[3759]	Speed: 68.475341 samples/sec	 accuracy=67.625000	 loss=1.318559	 lr=0.001000
Epoch[072] Batch [2349]/[3759]	Speed: 71.460203 samples/sec	 accuracy=67.619681	 loss=1.318701	 lr=0.001000
Epoch[072] Batch [2399]/[3759]	Speed: 69.507394 samples/sec	 accuracy=67.613932	 loss=1.318648	 lr=0.001000
Epoch[072] Batch [2449]/[3759]	Speed: 69.518108 samples/sec	 accuracy=67.606505	 loss=1.319215	 lr=0.001000
Epoch[072] Batch [2499]/[3759]	Speed: 69.492231 samples/sec	 accuracy=67.621250	 loss=1.318680	 lr=0.001000
Epoch[072] Batch [2549]/[3759]	Speed: 70.511457 samples/sec	 accuracy=67.631127	 loss=1.318330	 lr=0.001000
Epoch[072] Batch [2599]/[3759]	Speed: 69.783097 samples/sec	 accuracy=67.629808	 loss=1.318416	 lr=0.001000
Epoch[072] Batch [2649]/[3759]	Speed: 70.310867 samples/sec	 accuracy=67.616745	 loss=1.319227	 lr=0.001000
Epoch[072] Batch [2699]/[3759]	Speed: 68.636692 samples/sec	 accuracy=67.615741	 loss=1.319077	 lr=0.001000
Epoch[072] Batch [2749]/[3759]	Speed: 70.643391 samples/sec	 accuracy=67.621023	 loss=1.318708	 lr=0.001000
Epoch[072] Batch [2799]/[3759]	Speed: 69.821142 samples/sec	 accuracy=67.647321	 loss=1.318551	 lr=0.001000
Epoch[072] Batch [2849]/[3759]	Speed: 69.688908 samples/sec	 accuracy=67.678180	 loss=1.317259	 lr=0.001000
Epoch[072] Batch [2899]/[3759]	Speed: 68.107760 samples/sec	 accuracy=67.682651	 loss=1.316859	 lr=0.001000
Epoch[072] Batch [2949]/[3759]	Speed: 70.656731 samples/sec	 accuracy=67.685381	 loss=1.316121	 lr=0.001000
Epoch[072] Batch [2999]/[3759]	Speed: 70.425219 samples/sec	 accuracy=67.683854	 loss=1.316289	 lr=0.001000
Epoch[072] Batch [3049]/[3759]	Speed: 69.921582 samples/sec	 accuracy=67.689549	 loss=1.315455	 lr=0.001000
Epoch[072] Batch [3099]/[3759]	Speed: 68.595014 samples/sec	 accuracy=67.707157	 loss=1.314658	 lr=0.001000
Epoch[072] Batch [3149]/[3759]	Speed: 69.611114 samples/sec	 accuracy=67.686508	 loss=1.315300	 lr=0.001000
Epoch[072] Batch [3199]/[3759]	Speed: 69.922000 samples/sec	 accuracy=67.707520	 loss=1.314535	 lr=0.001000
Epoch[072] Batch [3249]/[3759]	Speed: 69.902041 samples/sec	 accuracy=67.708654	 loss=1.314235	 lr=0.001000
Epoch[072] Batch [3299]/[3759]	Speed: 69.962213 samples/sec	 accuracy=67.704545	 loss=1.314214	 lr=0.001000
Epoch[072] Batch [3349]/[3759]	Speed: 68.573247 samples/sec	 accuracy=67.698228	 loss=1.313990	 lr=0.001000
Epoch[072] Batch [3399]/[3759]	Speed: 71.246323 samples/sec	 accuracy=67.713235	 loss=1.313515	 lr=0.001000
Epoch[072] Batch [3449]/[3759]	Speed: 69.702353 samples/sec	 accuracy=67.709692	 loss=1.313635	 lr=0.001000
Epoch[072] Batch [3499]/[3759]	Speed: 67.895493 samples/sec	 accuracy=67.711607	 loss=1.313450	 lr=0.001000
Epoch[072] Batch [3549]/[3759]	Speed: 70.035999 samples/sec	 accuracy=67.696303	 loss=1.313412	 lr=0.001000
Epoch[072] Batch [3599]/[3759]	Speed: 71.680801 samples/sec	 accuracy=67.705295	 loss=1.313210	 lr=0.001000
Epoch[072] Batch [3649]/[3759]	Speed: 70.346182 samples/sec	 accuracy=67.688784	 loss=1.313861	 lr=0.001000
Epoch[072] Batch [3699]/[3759]	Speed: 69.452913 samples/sec	 accuracy=67.686655	 loss=1.313716	 lr=0.001000
Epoch[072] Batch [3749]/[3759]	Speed: 77.041044 samples/sec	 accuracy=67.691667	 loss=1.313462	 lr=0.001000
Batch [0049]/[0303]: acc-top1=61.812500 acc-top5=83.562500
Batch [0099]/[0303]: acc-top1=62.062500 acc-top5=83.421875
Batch [0149]/[0303]: acc-top1=62.833333 acc-top5=83.885417
Batch [0199]/[0303]: acc-top1=63.109375 acc-top5=84.320312
Batch [0249]/[0303]: acc-top1=63.368750 acc-top5=84.562500
Batch [0299]/[0303]: acc-top1=63.255208 acc-top5=84.432292
[Epoch 072] training: accuracy=67.688381	 loss=1.313476
[Epoch 072] speed: 69 samples/sec	time cost: 3747.749931
[Epoch 072] validation: acc-top1=63.242574 acc-top5=84.447195 loss=1.598622
Epoch[073] Batch [0049]/[3760]	Speed: 48.659017 samples/sec	 accuracy=68.281250	 loss=1.258486	 lr=0.001000
Epoch[073] Batch [0099]/[3760]	Speed: 66.271766 samples/sec	 accuracy=68.578125	 loss=1.253000	 lr=0.001000
Epoch[073] Batch [0149]/[3760]	Speed: 71.295557 samples/sec	 accuracy=68.291667	 loss=1.275645	 lr=0.001000
Epoch[073] Batch [0199]/[3760]	Speed: 69.395352 samples/sec	 accuracy=68.148438	 loss=1.276734	 lr=0.001000
Epoch[073] Batch [0249]/[3760]	Speed: 70.616006 samples/sec	 accuracy=68.550000	 loss=1.269153	 lr=0.001000
Epoch[073] Batch [0299]/[3760]	Speed: 68.379615 samples/sec	 accuracy=68.276042	 loss=1.281278	 lr=0.001000
Epoch[073] Batch [0349]/[3760]	Speed: 69.661653 samples/sec	 accuracy=68.290179	 loss=1.283713	 lr=0.001000
Epoch[073] Batch [0399]/[3760]	Speed: 70.667371 samples/sec	 accuracy=68.183594	 loss=1.288524	 lr=0.001000
Epoch[073] Batch [0449]/[3760]	Speed: 69.014605 samples/sec	 accuracy=68.225694	 loss=1.287823	 lr=0.001000
Epoch[073] Batch [0499]/[3760]	Speed: 69.650940 samples/sec	 accuracy=68.293750	 loss=1.285148	 lr=0.001000
Epoch[073] Batch [0549]/[3760]	Speed: 70.414711 samples/sec	 accuracy=68.335227	 loss=1.283656	 lr=0.001000
Epoch[073] Batch [0599]/[3760]	Speed: 66.594718 samples/sec	 accuracy=68.338542	 loss=1.286854	 lr=0.001000
Epoch[073] Batch [0649]/[3760]	Speed: 71.401005 samples/sec	 accuracy=68.326923	 loss=1.288462	 lr=0.001000
Epoch[073] Batch [0699]/[3760]	Speed: 71.166615 samples/sec	 accuracy=68.301339	 loss=1.289692	 lr=0.001000
Epoch[073] Batch [0749]/[3760]	Speed: 68.540459 samples/sec	 accuracy=68.335417	 loss=1.288790	 lr=0.001000
Epoch[073] Batch [0799]/[3760]	Speed: 69.986786 samples/sec	 accuracy=68.308594	 loss=1.289719	 lr=0.001000
Epoch[073] Batch [0849]/[3760]	Speed: 70.651077 samples/sec	 accuracy=68.336397	 loss=1.287623	 lr=0.001000
Epoch[073] Batch [0899]/[3760]	Speed: 69.311886 samples/sec	 accuracy=68.284722	 loss=1.290571	 lr=0.001000
Epoch[073] Batch [0949]/[3760]	Speed: 70.382805 samples/sec	 accuracy=68.289474	 loss=1.290889	 lr=0.001000
Epoch[073] Batch [0999]/[3760]	Speed: 68.593709 samples/sec	 accuracy=68.234375	 loss=1.289900	 lr=0.001000
Epoch[073] Batch [1049]/[3760]	Speed: 70.214694 samples/sec	 accuracy=68.157738	 loss=1.293761	 lr=0.001000
Epoch[073] Batch [1099]/[3760]	Speed: 69.839950 samples/sec	 accuracy=68.156250	 loss=1.292680	 lr=0.001000
Epoch[073] Batch [1149]/[3760]	Speed: 70.630276 samples/sec	 accuracy=68.146739	 loss=1.292471	 lr=0.001000
Epoch[073] Batch [1199]/[3760]	Speed: 69.875036 samples/sec	 accuracy=68.122396	 loss=1.292792	 lr=0.001000
Epoch[073] Batch [1249]/[3760]	Speed: 70.293670 samples/sec	 accuracy=68.158750	 loss=1.291478	 lr=0.001000
Epoch[073] Batch [1299]/[3760]	Speed: 69.632849 samples/sec	 accuracy=68.126202	 loss=1.292687	 lr=0.001000
Epoch[073] Batch [1349]/[3760]	Speed: 68.359885 samples/sec	 accuracy=68.111111	 loss=1.293745	 lr=0.001000
Epoch[073] Batch [1399]/[3760]	Speed: 71.156441 samples/sec	 accuracy=68.060268	 loss=1.294781	 lr=0.001000
Epoch[073] Batch [1449]/[3760]	Speed: 70.263204 samples/sec	 accuracy=67.992457	 loss=1.297314	 lr=0.001000
Epoch[073] Batch [1499]/[3760]	Speed: 69.124356 samples/sec	 accuracy=67.993750	 loss=1.297910	 lr=0.001000
Epoch[073] Batch [1549]/[3760]	Speed: 70.636603 samples/sec	 accuracy=68.031250	 loss=1.296791	 lr=0.001000
Epoch[073] Batch [1599]/[3760]	Speed: 69.184476 samples/sec	 accuracy=68.044922	 loss=1.296971	 lr=0.001000
Epoch[073] Batch [1649]/[3760]	Speed: 68.121850 samples/sec	 accuracy=68.020833	 loss=1.297007	 lr=0.001000
Epoch[073] Batch [1699]/[3760]	Speed: 70.938412 samples/sec	 accuracy=67.978860	 loss=1.299306	 lr=0.001000
Epoch[073] Batch [1749]/[3760]	Speed: 69.106350 samples/sec	 accuracy=67.955357	 loss=1.299630	 lr=0.001000
Epoch[073] Batch [1799]/[3760]	Speed: 71.210459 samples/sec	 accuracy=67.944444	 loss=1.298430	 lr=0.001000
Epoch[073] Batch [1849]/[3760]	Speed: 70.248857 samples/sec	 accuracy=67.928209	 loss=1.298710	 lr=0.001000
Epoch[073] Batch [1899]/[3760]	Speed: 69.787069 samples/sec	 accuracy=67.940789	 loss=1.299007	 lr=0.001000
Epoch[073] Batch [1949]/[3760]	Speed: 68.842517 samples/sec	 accuracy=67.926282	 loss=1.299618	 lr=0.001000
Epoch[073] Batch [1999]/[3760]	Speed: 70.250414 samples/sec	 accuracy=67.937500	 loss=1.299694	 lr=0.001000
Epoch[073] Batch [2049]/[3760]	Speed: 69.656335 samples/sec	 accuracy=67.939024	 loss=1.300112	 lr=0.001000
Epoch[073] Batch [2099]/[3760]	Speed: 68.213809 samples/sec	 accuracy=67.916667	 loss=1.300514	 lr=0.001000
Epoch[073] Batch [2149]/[3760]	Speed: 71.046165 samples/sec	 accuracy=67.898256	 loss=1.300619	 lr=0.001000
Epoch[073] Batch [2199]/[3760]	Speed: 68.991318 samples/sec	 accuracy=67.898438	 loss=1.300228	 lr=0.001000
Epoch[073] Batch [2249]/[3760]	Speed: 71.457916 samples/sec	 accuracy=67.894444	 loss=1.300244	 lr=0.001000
Epoch[073] Batch [2299]/[3760]	Speed: 70.524428 samples/sec	 accuracy=67.867527	 loss=1.300688	 lr=0.001000
Epoch[073] Batch [2349]/[3760]	Speed: 68.894096 samples/sec	 accuracy=67.870346	 loss=1.300928	 lr=0.001000
Epoch[073] Batch [2399]/[3760]	Speed: 69.780182 samples/sec	 accuracy=67.854818	 loss=1.302178	 lr=0.001000
Epoch[073] Batch [2449]/[3760]	Speed: 68.480283 samples/sec	 accuracy=67.866709	 loss=1.302285	 lr=0.001000
Epoch[073] Batch [2499]/[3760]	Speed: 70.127505 samples/sec	 accuracy=67.848125	 loss=1.302758	 lr=0.001000
Epoch[073] Batch [2549]/[3760]	Speed: 68.797790 samples/sec	 accuracy=67.873162	 loss=1.302849	 lr=0.001000
Epoch[073] Batch [2599]/[3760]	Speed: 69.390099 samples/sec	 accuracy=67.884615	 loss=1.301989	 lr=0.001000
Epoch[073] Batch [2649]/[3760]	Speed: 70.653443 samples/sec	 accuracy=67.902712	 loss=1.302118	 lr=0.001000
Epoch[073] Batch [2699]/[3760]	Speed: 69.965424 samples/sec	 accuracy=67.886574	 loss=1.302115	 lr=0.001000
Epoch[073] Batch [2749]/[3760]	Speed: 68.837560 samples/sec	 accuracy=67.872727	 loss=1.302925	 lr=0.001000
Epoch[073] Batch [2799]/[3760]	Speed: 70.128392 samples/sec	 accuracy=67.875558	 loss=1.303083	 lr=0.001000
Epoch[073] Batch [2849]/[3760]	Speed: 69.054145 samples/sec	 accuracy=67.878838	 loss=1.303044	 lr=0.001000
Epoch[073] Batch [2899]/[3760]	Speed: 70.492653 samples/sec	 accuracy=67.885776	 loss=1.303151	 lr=0.001000
Epoch[073] Batch [2949]/[3760]	Speed: 70.052449 samples/sec	 accuracy=67.873411	 loss=1.302918	 lr=0.001000
Epoch[073] Batch [2999]/[3760]	Speed: 69.044528 samples/sec	 accuracy=67.867708	 loss=1.302842	 lr=0.001000
Epoch[073] Batch [3049]/[3760]	Speed: 70.158195 samples/sec	 accuracy=67.855020	 loss=1.303627	 lr=0.001000
Epoch[073] Batch [3099]/[3760]	Speed: 70.128212 samples/sec	 accuracy=67.869456	 loss=1.303622	 lr=0.001000
Epoch[073] Batch [3149]/[3760]	Speed: 70.095022 samples/sec	 accuracy=67.861111	 loss=1.303658	 lr=0.001000
Epoch[073] Batch [3199]/[3760]	Speed: 69.042898 samples/sec	 accuracy=67.868164	 loss=1.303572	 lr=0.001000
Epoch[073] Batch [3249]/[3760]	Speed: 69.913020 samples/sec	 accuracy=67.864904	 loss=1.304259	 lr=0.001000
Epoch[073] Batch [3299]/[3760]	Speed: 70.107456 samples/sec	 accuracy=67.842803	 loss=1.305135	 lr=0.001000
Epoch[073] Batch [3349]/[3760]	Speed: 69.358598 samples/sec	 accuracy=67.806903	 loss=1.306331	 lr=0.001000
Epoch[073] Batch [3399]/[3760]	Speed: 66.733903 samples/sec	 accuracy=67.812500	 loss=1.306011	 lr=0.001000
Epoch[073] Batch [3449]/[3760]	Speed: 73.235321 samples/sec	 accuracy=67.808877	 loss=1.306026	 lr=0.001000
Epoch[073] Batch [3499]/[3760]	Speed: 70.259833 samples/sec	 accuracy=67.816964	 loss=1.305489	 lr=0.001000
Epoch[073] Batch [3549]/[3760]	Speed: 69.001759 samples/sec	 accuracy=67.809419	 loss=1.305640	 lr=0.001000
Epoch[073] Batch [3599]/[3760]	Speed: 70.018031 samples/sec	 accuracy=67.799913	 loss=1.305834	 lr=0.001000
Epoch[073] Batch [3649]/[3760]	Speed: 70.080054 samples/sec	 accuracy=67.790240	 loss=1.306050	 lr=0.001000
Epoch[073] Batch [3699]/[3760]	Speed: 69.028281 samples/sec	 accuracy=67.789696	 loss=1.306487	 lr=0.001000
Epoch[073] Batch [3749]/[3760]	Speed: 76.975165 samples/sec	 accuracy=67.785833	 loss=1.306429	 lr=0.001000
Batch [0049]/[0303]: acc-top1=61.656250 acc-top5=83.593750
Batch [0099]/[0303]: acc-top1=61.906250 acc-top5=83.468750
Batch [0149]/[0303]: acc-top1=62.802083 acc-top5=84.000000
Batch [0199]/[0303]: acc-top1=63.140625 acc-top5=84.359375
Batch [0249]/[0303]: acc-top1=63.400000 acc-top5=84.625000
Batch [0299]/[0303]: acc-top1=63.197917 acc-top5=84.546875
[Epoch 073] training: accuracy=67.785904	 loss=1.306474
[Epoch 073] speed: 69 samples/sec	time cost: 3752.810707
[Epoch 073] validation: acc-top1=63.154909 acc-top5=84.550330 loss=1.595929
Epoch[074] Batch [0049]/[3760]	Speed: 48.783881 samples/sec	 accuracy=68.218750	 loss=1.315880	 lr=0.001000
Epoch[074] Batch [0099]/[3760]	Speed: 67.904105 samples/sec	 accuracy=68.234375	 loss=1.314083	 lr=0.001000
Epoch[074] Batch [0149]/[3760]	Speed: 70.226196 samples/sec	 accuracy=68.010417	 loss=1.318570	 lr=0.001000
Epoch[074] Batch [0199]/[3760]	Speed: 70.576642 samples/sec	 accuracy=68.414062	 loss=1.298204	 lr=0.001000
Epoch[074] Batch [0249]/[3760]	Speed: 68.911156 samples/sec	 accuracy=68.225000	 loss=1.305097	 lr=0.001000
Epoch[074] Batch [0299]/[3760]	Speed: 70.338711 samples/sec	 accuracy=68.296875	 loss=1.299990	 lr=0.001000
Epoch[074] Batch [0349]/[3760]	Speed: 69.508960 samples/sec	 accuracy=68.089286	 loss=1.302980	 lr=0.001000
Epoch[074] Batch [0399]/[3760]	Speed: 69.546829 samples/sec	 accuracy=68.011719	 loss=1.308854	 lr=0.001000
Epoch[074] Batch [0449]/[3760]	Speed: 69.582947 samples/sec	 accuracy=68.034722	 loss=1.308218	 lr=0.001000
Epoch[074] Batch [0499]/[3760]	Speed: 67.901188 samples/sec	 accuracy=67.962500	 loss=1.307047	 lr=0.001000
Epoch[074] Batch [0549]/[3760]	Speed: 70.198487 samples/sec	 accuracy=67.982955	 loss=1.304064	 lr=0.001000
Epoch[074] Batch [0599]/[3760]	Speed: 67.102916 samples/sec	 accuracy=67.950521	 loss=1.302631	 lr=0.001000
Epoch[074] Batch [0649]/[3760]	Speed: 72.133126 samples/sec	 accuracy=68.024038	 loss=1.301303	 lr=0.001000
Epoch[074] Batch [0699]/[3760]	Speed: 69.308210 samples/sec	 accuracy=68.008929	 loss=1.300584	 lr=0.001000
Epoch[074] Batch [0749]/[3760]	Speed: 70.318020 samples/sec	 accuracy=68.060417	 loss=1.300293	 lr=0.001000
Epoch[074] Batch [0799]/[3760]	Speed: 70.821559 samples/sec	 accuracy=68.025391	 loss=1.301686	 lr=0.001000
Epoch[074] Batch [0849]/[3760]	Speed: 69.579149 samples/sec	 accuracy=67.926471	 loss=1.304833	 lr=0.001000
Epoch[074] Batch [0899]/[3760]	Speed: 66.852284 samples/sec	 accuracy=67.921875	 loss=1.304217	 lr=0.001000
Epoch[074] Batch [0949]/[3760]	Speed: 71.607961 samples/sec	 accuracy=67.955592	 loss=1.302192	 lr=0.001000
Epoch[074] Batch [0999]/[3760]	Speed: 70.903955 samples/sec	 accuracy=67.939063	 loss=1.302368	 lr=0.001000
Epoch[074] Batch [1049]/[3760]	Speed: 69.677368 samples/sec	 accuracy=67.944940	 loss=1.299349	 lr=0.001000
Epoch[074] Batch [1099]/[3760]	Speed: 69.392379 samples/sec	 accuracy=67.941761	 loss=1.298304	 lr=0.001000
Epoch[074] Batch [1149]/[3760]	Speed: 68.652237 samples/sec	 accuracy=67.902174	 loss=1.298808	 lr=0.001000
Epoch[074] Batch [1199]/[3760]	Speed: 69.172380 samples/sec	 accuracy=67.897135	 loss=1.299326	 lr=0.001000
Epoch[074] Batch [1249]/[3760]	Speed: 70.995457 samples/sec	 accuracy=67.892500	 loss=1.298975	 lr=0.001000
Epoch[074] Batch [1299]/[3760]	Speed: 70.349575 samples/sec	 accuracy=67.923077	 loss=1.298042	 lr=0.001000
Epoch[074] Batch [1349]/[3760]	Speed: 69.119082 samples/sec	 accuracy=67.896991	 loss=1.300414	 lr=0.001000
Epoch[074] Batch [1399]/[3760]	Speed: 69.821717 samples/sec	 accuracy=67.907366	 loss=1.301997	 lr=0.001000
Epoch[074] Batch [1449]/[3760]	Speed: 70.128477 samples/sec	 accuracy=67.877155	 loss=1.302758	 lr=0.001000
Epoch[074] Batch [1499]/[3760]	Speed: 70.441351 samples/sec	 accuracy=67.907292	 loss=1.302150	 lr=0.001000
Epoch[074] Batch [1549]/[3760]	Speed: 68.075468 samples/sec	 accuracy=67.894153	 loss=1.302285	 lr=0.001000
Epoch[074] Batch [1599]/[3760]	Speed: 70.829747 samples/sec	 accuracy=67.933594	 loss=1.302485	 lr=0.001000
Epoch[074] Batch [1649]/[3760]	Speed: 68.209549 samples/sec	 accuracy=67.958333	 loss=1.300747	 lr=0.001000
Epoch[074] Batch [1699]/[3760]	Speed: 71.051769 samples/sec	 accuracy=67.970588	 loss=1.300989	 lr=0.001000
Epoch[074] Batch [1749]/[3760]	Speed: 71.420390 samples/sec	 accuracy=67.948214	 loss=1.301688	 lr=0.001000
Epoch[074] Batch [1799]/[3760]	Speed: 67.325299 samples/sec	 accuracy=67.930556	 loss=1.301670	 lr=0.001000
Epoch[074] Batch [1849]/[3760]	Speed: 70.739005 samples/sec	 accuracy=67.905405	 loss=1.302404	 lr=0.001000
Epoch[074] Batch [1899]/[3760]	Speed: 69.250019 samples/sec	 accuracy=67.944079	 loss=1.300625	 lr=0.001000
Epoch[074] Batch [1949]/[3760]	Speed: 69.683282 samples/sec	 accuracy=67.947917	 loss=1.300494	 lr=0.001000
Epoch[074] Batch [1999]/[3760]	Speed: 68.135594 samples/sec	 accuracy=67.923437	 loss=1.300666	 lr=0.001000
Epoch[074] Batch [2049]/[3760]	Speed: 71.363561 samples/sec	 accuracy=67.923018	 loss=1.300535	 lr=0.001000
Epoch[074] Batch [2099]/[3760]	Speed: 67.263666 samples/sec	 accuracy=67.928571	 loss=1.299932	 lr=0.001000
Epoch[074] Batch [2149]/[3760]	Speed: 69.589846 samples/sec	 accuracy=67.899709	 loss=1.300897	 lr=0.001000
Epoch[074] Batch [2199]/[3760]	Speed: 72.115356 samples/sec	 accuracy=67.906250	 loss=1.301149	 lr=0.001000
Epoch[074] Batch [2249]/[3760]	Speed: 69.134662 samples/sec	 accuracy=67.895139	 loss=1.301470	 lr=0.001000
Epoch[074] Batch [2299]/[3760]	Speed: 68.778247 samples/sec	 accuracy=67.920516	 loss=1.300529	 lr=0.001000
Epoch[074] Batch [2349]/[3760]	Speed: 71.564939 samples/sec	 accuracy=67.902926	 loss=1.301247	 lr=0.001000
Epoch[074] Batch [2399]/[3760]	Speed: 64.995898 samples/sec	 accuracy=67.919922	 loss=1.300735	 lr=0.001000
Epoch[074] Batch [2449]/[3760]	Speed: 74.433048 samples/sec	 accuracy=67.902423	 loss=1.301467	 lr=0.001000
Epoch[074] Batch [2499]/[3760]	Speed: 68.795342 samples/sec	 accuracy=67.885625	 loss=1.301693	 lr=0.001000
Epoch[074] Batch [2549]/[3760]	Speed: 66.405643 samples/sec	 accuracy=67.866422	 loss=1.301784	 lr=0.001000
Epoch[074] Batch [2599]/[3760]	Speed: 72.919076 samples/sec	 accuracy=67.861178	 loss=1.302179	 lr=0.001000
Epoch[074] Batch [2649]/[3760]	Speed: 70.327491 samples/sec	 accuracy=67.866745	 loss=1.302407	 lr=0.001000
Epoch[074] Batch [2699]/[3760]	Speed: 69.257973 samples/sec	 accuracy=67.839699	 loss=1.303562	 lr=0.001000
Epoch[074] Batch [2749]/[3760]	Speed: 71.213134 samples/sec	 accuracy=67.855114	 loss=1.303221	 lr=0.001000
Epoch[074] Batch [2799]/[3760]	Speed: 68.276741 samples/sec	 accuracy=67.885045	 loss=1.301943	 lr=0.001000
Epoch[074] Batch [2849]/[3760]	Speed: 70.765842 samples/sec	 accuracy=67.885965	 loss=1.301889	 lr=0.001000
Epoch[074] Batch [2899]/[3760]	Speed: 69.546977 samples/sec	 accuracy=67.875539	 loss=1.302691	 lr=0.001000
Epoch[074] Batch [2949]/[3760]	Speed: 69.346848 samples/sec	 accuracy=67.884004	 loss=1.302599	 lr=0.001000
Epoch[074] Batch [2999]/[3760]	Speed: 69.565032 samples/sec	 accuracy=67.904167	 loss=1.302049	 lr=0.001000
Epoch[074] Batch [3049]/[3760]	Speed: 69.983074 samples/sec	 accuracy=67.920594	 loss=1.301390	 lr=0.001000
Epoch[074] Batch [3099]/[3760]	Speed: 68.333903 samples/sec	 accuracy=67.919859	 loss=1.301096	 lr=0.001000
Epoch[074] Batch [3149]/[3760]	Speed: 71.454507 samples/sec	 accuracy=67.909722	 loss=1.301360	 lr=0.001000
Epoch[074] Batch [3199]/[3760]	Speed: 68.903847 samples/sec	 accuracy=67.923828	 loss=1.301133	 lr=0.001000
Epoch[074] Batch [3249]/[3760]	Speed: 69.322006 samples/sec	 accuracy=67.933173	 loss=1.300434	 lr=0.001000
Epoch[074] Batch [3299]/[3760]	Speed: 69.683574 samples/sec	 accuracy=67.915246	 loss=1.300765	 lr=0.001000
Epoch[074] Batch [3349]/[3760]	Speed: 70.634355 samples/sec	 accuracy=67.922108	 loss=1.301069	 lr=0.001000
Epoch[074] Batch [3399]/[3760]	Speed: 68.382644 samples/sec	 accuracy=67.905331	 loss=1.301774	 lr=0.001000
Epoch[074] Batch [3449]/[3760]	Speed: 69.560990 samples/sec	 accuracy=67.888587	 loss=1.302568	 lr=0.001000
Epoch[074] Batch [3499]/[3760]	Speed: 68.459884 samples/sec	 accuracy=67.903571	 loss=1.302180	 lr=0.001000
Epoch[074] Batch [3549]/[3760]	Speed: 69.403139 samples/sec	 accuracy=67.906690	 loss=1.302316	 lr=0.001000
Epoch[074] Batch [3599]/[3760]	Speed: 68.995265 samples/sec	 accuracy=67.915799	 loss=1.301790	 lr=0.001000
Epoch[074] Batch [3649]/[3760]	Speed: 67.781350 samples/sec	 accuracy=67.898545	 loss=1.301785	 lr=0.001000
Epoch[074] Batch [3699]/[3760]	Speed: 72.590455 samples/sec	 accuracy=67.891047	 loss=1.302340	 lr=0.001000
Epoch[074] Batch [3749]/[3760]	Speed: 76.959440 samples/sec	 accuracy=67.890000	 loss=1.302651	 lr=0.001000
Batch [0049]/[0303]: acc-top1=61.875000 acc-top5=83.125000
Batch [0099]/[0303]: acc-top1=61.953125 acc-top5=83.312500
Batch [0149]/[0303]: acc-top1=63.114583 acc-top5=83.791667
Batch [0199]/[0303]: acc-top1=63.562500 acc-top5=84.257812
Batch [0249]/[0303]: acc-top1=63.768750 acc-top5=84.493750
Batch [0299]/[0303]: acc-top1=63.656250 acc-top5=84.380208
[Epoch 074] training: accuracy=67.885223	 loss=1.302734
[Epoch 074] speed: 69 samples/sec	time cost: 3754.857901
[Epoch 074] validation: acc-top1=63.634488 acc-top5=84.390470 loss=1.596705
Epoch[075] Batch [0049]/[3759]	Speed: 48.882557 samples/sec	 accuracy=69.218750	 loss=1.252457	 lr=0.001000
Epoch[075] Batch [0099]/[3759]	Speed: 67.523694 samples/sec	 accuracy=68.000000	 loss=1.278513	 lr=0.001000
Epoch[075] Batch [0149]/[3759]	Speed: 70.147160 samples/sec	 accuracy=67.927083	 loss=1.286197	 lr=0.001000
Epoch[075] Batch [0199]/[3759]	Speed: 68.875866 samples/sec	 accuracy=68.078125	 loss=1.286926	 lr=0.001000
Epoch[075] Batch [0249]/[3759]	Speed: 69.342048 samples/sec	 accuracy=68.212500	 loss=1.287532	 lr=0.001000
Epoch[075] Batch [0299]/[3759]	Speed: 71.170259 samples/sec	 accuracy=68.369792	 loss=1.288601	 lr=0.001000
Epoch[075] Batch [0349]/[3759]	Speed: 69.424608 samples/sec	 accuracy=68.446429	 loss=1.285450	 lr=0.001000
Epoch[075] Batch [0399]/[3759]	Speed: 70.800154 samples/sec	 accuracy=68.285156	 loss=1.290084	 lr=0.001000
Epoch[075] Batch [0449]/[3759]	Speed: 68.358998 samples/sec	 accuracy=68.263889	 loss=1.290957	 lr=0.001000
Epoch[075] Batch [0499]/[3759]	Speed: 71.467014 samples/sec	 accuracy=68.228125	 loss=1.291654	 lr=0.001000
Epoch[075] Batch [0549]/[3759]	Speed: 69.049722 samples/sec	 accuracy=68.258523	 loss=1.290589	 lr=0.001000
Epoch[075] Batch [0599]/[3759]	Speed: 71.111344 samples/sec	 accuracy=68.221354	 loss=1.291445	 lr=0.001000
Epoch[075] Batch [0649]/[3759]	Speed: 69.160018 samples/sec	 accuracy=68.206731	 loss=1.291630	 lr=0.001000
Epoch[075] Batch [0699]/[3759]	Speed: 68.518778 samples/sec	 accuracy=68.060268	 loss=1.294589	 lr=0.001000
Epoch[075] Batch [0749]/[3759]	Speed: 70.229877 samples/sec	 accuracy=68.114583	 loss=1.291323	 lr=0.001000
Epoch[075] Batch [0799]/[3759]	Speed: 70.639490 samples/sec	 accuracy=68.150391	 loss=1.288928	 lr=0.001000
Epoch[075] Batch [0849]/[3759]	Speed: 70.151522 samples/sec	 accuracy=68.178309	 loss=1.287505	 lr=0.001000
Epoch[075] Batch [0899]/[3759]	Speed: 68.437640 samples/sec	 accuracy=68.154514	 loss=1.288393	 lr=0.001000
Epoch[075] Batch [0949]/[3759]	Speed: 70.685825 samples/sec	 accuracy=68.100329	 loss=1.290533	 lr=0.001000
Epoch[075] Batch [0999]/[3759]	Speed: 70.579153 samples/sec	 accuracy=68.192188	 loss=1.288446	 lr=0.001000
Epoch[075] Batch [1049]/[3759]	Speed: 67.125760 samples/sec	 accuracy=68.217262	 loss=1.288544	 lr=0.001000
Epoch[075] Batch [1099]/[3759]	Speed: 70.894270 samples/sec	 accuracy=68.177557	 loss=1.289795	 lr=0.001000
Epoch[075] Batch [1149]/[3759]	Speed: 70.768442 samples/sec	 accuracy=68.222826	 loss=1.288204	 lr=0.001000
Epoch[075] Batch [1199]/[3759]	Speed: 69.792064 samples/sec	 accuracy=68.252604	 loss=1.287555	 lr=0.001000
Epoch[075] Batch [1249]/[3759]	Speed: 68.865969 samples/sec	 accuracy=68.227500	 loss=1.288341	 lr=0.001000
Epoch[075] Batch [1299]/[3759]	Speed: 70.318876 samples/sec	 accuracy=68.241587	 loss=1.288237	 lr=0.001000
Epoch[075] Batch [1349]/[3759]	Speed: 70.025958 samples/sec	 accuracy=68.200231	 loss=1.288882	 lr=0.001000
Epoch[075] Batch [1399]/[3759]	Speed: 69.247202 samples/sec	 accuracy=68.165179	 loss=1.290450	 lr=0.001000
Epoch[075] Batch [1449]/[3759]	Speed: 70.702173 samples/sec	 accuracy=68.181034	 loss=1.291128	 lr=0.001000
Epoch[075] Batch [1499]/[3759]	Speed: 67.866793 samples/sec	 accuracy=68.181250	 loss=1.289833	 lr=0.001000
Epoch[075] Batch [1549]/[3759]	Speed: 71.250606 samples/sec	 accuracy=68.183468	 loss=1.290297	 lr=0.001000
Epoch[075] Batch [1599]/[3759]	Speed: 70.214339 samples/sec	 accuracy=68.222656	 loss=1.289832	 lr=0.001000
Epoch[075] Batch [1649]/[3759]	Speed: 70.583400 samples/sec	 accuracy=68.214962	 loss=1.290046	 lr=0.001000
Epoch[075] Batch [1699]/[3759]	Speed: 68.281095 samples/sec	 accuracy=68.199449	 loss=1.291242	 lr=0.001000
Epoch[075] Batch [1749]/[3759]	Speed: 72.339498 samples/sec	 accuracy=68.192857	 loss=1.291333	 lr=0.001000
Epoch[075] Batch [1799]/[3759]	Speed: 68.639583 samples/sec	 accuracy=68.176215	 loss=1.291569	 lr=0.001000
Epoch[075] Batch [1849]/[3759]	Speed: 69.714999 samples/sec	 accuracy=68.156250	 loss=1.291873	 lr=0.001000
Epoch[075] Batch [1899]/[3759]	Speed: 70.932033 samples/sec	 accuracy=68.130757	 loss=1.292710	 lr=0.001000
Epoch[075] Batch [1949]/[3759]	Speed: 68.801071 samples/sec	 accuracy=68.114583	 loss=1.293773	 lr=0.001000
Epoch[075] Batch [1999]/[3759]	Speed: 70.605185 samples/sec	 accuracy=68.114063	 loss=1.293369	 lr=0.001000
Epoch[075] Batch [2049]/[3759]	Speed: 69.459221 samples/sec	 accuracy=68.108994	 loss=1.294306	 lr=0.001000
Epoch[075] Batch [2099]/[3759]	Speed: 69.959721 samples/sec	 accuracy=68.119792	 loss=1.294401	 lr=0.001000
Epoch[075] Batch [2149]/[3759]	Speed: 68.665640 samples/sec	 accuracy=68.100291	 loss=1.294790	 lr=0.001000
Epoch[075] Batch [2199]/[3759]	Speed: 71.275025 samples/sec	 accuracy=68.106534	 loss=1.295166	 lr=0.001000
Epoch[075] Batch [2249]/[3759]	Speed: 69.453544 samples/sec	 accuracy=68.101389	 loss=1.295637	 lr=0.001000
Epoch[075] Batch [2299]/[3759]	Speed: 69.633120 samples/sec	 accuracy=68.099185	 loss=1.295937	 lr=0.001000
Epoch[075] Batch [2349]/[3759]	Speed: 69.710413 samples/sec	 accuracy=68.083112	 loss=1.296148	 lr=0.001000
Epoch[075] Batch [2399]/[3759]	Speed: 69.159783 samples/sec	 accuracy=68.054036	 loss=1.297227	 lr=0.001000
Epoch[075] Batch [2449]/[3759]	Speed: 69.839771 samples/sec	 accuracy=68.066327	 loss=1.296861	 lr=0.001000
Epoch[075] Batch [2499]/[3759]	Speed: 70.181530 samples/sec	 accuracy=68.051875	 loss=1.297592	 lr=0.001000
Epoch[075] Batch [2549]/[3759]	Speed: 70.085773 samples/sec	 accuracy=68.055760	 loss=1.297259	 lr=0.001000
Epoch[075] Batch [2599]/[3759]	Speed: 70.533609 samples/sec	 accuracy=68.086538	 loss=1.296473	 lr=0.001000
Epoch[075] Batch [2649]/[3759]	Speed: 69.881431 samples/sec	 accuracy=68.070755	 loss=1.296679	 lr=0.001000
Epoch[075] Batch [2699]/[3759]	Speed: 69.981705 samples/sec	 accuracy=68.053241	 loss=1.297568	 lr=0.001000
Epoch[075] Batch [2749]/[3759]	Speed: 68.894563 samples/sec	 accuracy=68.036932	 loss=1.297632	 lr=0.001000
Epoch[075] Batch [2799]/[3759]	Speed: 70.779534 samples/sec	 accuracy=68.017857	 loss=1.297677	 lr=0.001000
Epoch[075] Batch [2849]/[3759]	Speed: 69.727571 samples/sec	 accuracy=68.026864	 loss=1.297402	 lr=0.001000
Epoch[075] Batch [2899]/[3759]	Speed: 69.819013 samples/sec	 accuracy=68.042565	 loss=1.296983	 lr=0.001000
Epoch[075] Batch [2949]/[3759]	Speed: 69.745665 samples/sec	 accuracy=68.024894	 loss=1.297008	 lr=0.001000
Epoch[075] Batch [2999]/[3759]	Speed: 70.072999 samples/sec	 accuracy=68.017187	 loss=1.297275	 lr=0.001000
Epoch[075] Batch [3049]/[3759]	Speed: 70.163304 samples/sec	 accuracy=68.013832	 loss=1.297043	 lr=0.001000
Epoch[075] Batch [3099]/[3759]	Speed: 69.574719 samples/sec	 accuracy=67.986391	 loss=1.297842	 lr=0.001000
Epoch[075] Batch [3149]/[3759]	Speed: 71.132250 samples/sec	 accuracy=67.980655	 loss=1.298439	 lr=0.001000
Epoch[075] Batch [3199]/[3759]	Speed: 68.430922 samples/sec	 accuracy=67.979980	 loss=1.298560	 lr=0.001000
Epoch[075] Batch [3249]/[3759]	Speed: 69.954319 samples/sec	 accuracy=67.996154	 loss=1.298023	 lr=0.001000
Epoch[075] Batch [3299]/[3759]	Speed: 70.882402 samples/sec	 accuracy=67.997633	 loss=1.297650	 lr=0.001000
Epoch[075] Batch [3349]/[3759]	Speed: 68.904372 samples/sec	 accuracy=67.974813	 loss=1.298426	 lr=0.001000
Epoch[075] Batch [3399]/[3759]	Speed: 70.429144 samples/sec	 accuracy=67.956801	 loss=1.299542	 lr=0.001000
Epoch[075] Batch [3449]/[3759]	Speed: 68.905698 samples/sec	 accuracy=67.955616	 loss=1.299306	 lr=0.001000
Epoch[075] Batch [3499]/[3759]	Speed: 69.457716 samples/sec	 accuracy=67.943304	 loss=1.299575	 lr=0.001000
Epoch[075] Batch [3549]/[3759]	Speed: 68.487104 samples/sec	 accuracy=67.921215	 loss=1.300245	 lr=0.001000
Epoch[075] Batch [3599]/[3759]	Speed: 71.084816 samples/sec	 accuracy=67.901910	 loss=1.301163	 lr=0.001000
Epoch[075] Batch [3649]/[3759]	Speed: 67.179426 samples/sec	 accuracy=67.918664	 loss=1.300704	 lr=0.001000
Epoch[075] Batch [3699]/[3759]	Speed: 71.309280 samples/sec	 accuracy=67.935389	 loss=1.300456	 lr=0.001000
Epoch[075] Batch [3749]/[3759]	Speed: 77.052079 samples/sec	 accuracy=67.920833	 loss=1.300853	 lr=0.001000
Batch [0049]/[0303]: acc-top1=62.625000 acc-top5=83.500000
Batch [0099]/[0303]: acc-top1=62.625000 acc-top5=83.546875
Batch [0149]/[0303]: acc-top1=63.677083 acc-top5=84.052083
Batch [0199]/[0303]: acc-top1=64.015625 acc-top5=84.468750
Batch [0249]/[0303]: acc-top1=64.081250 acc-top5=84.643750
Batch [0299]/[0303]: acc-top1=63.692708 acc-top5=84.557292
[Epoch 075] training: accuracy=67.917831	 loss=1.300834
[Epoch 075] speed: 69 samples/sec	time cost: 3746.972473
[Epoch 075] validation: acc-top1=63.665429 acc-top5=84.570957 loss=1.585557
Epoch[076] Batch [0049]/[3760]	Speed: 47.318829 samples/sec	 accuracy=69.218750	 loss=1.277848	 lr=0.001000
Epoch[076] Batch [0099]/[3760]	Speed: 70.662460 samples/sec	 accuracy=68.671875	 loss=1.276659	 lr=0.001000
Epoch[076] Batch [0149]/[3760]	Speed: 68.395176 samples/sec	 accuracy=68.572917	 loss=1.276267	 lr=0.001000
Epoch[076] Batch [0199]/[3760]	Speed: 69.688918 samples/sec	 accuracy=68.648438	 loss=1.275857	 lr=0.001000
Epoch[076] Batch [0249]/[3760]	Speed: 70.779151 samples/sec	 accuracy=68.481250	 loss=1.286467	 lr=0.001000
Epoch[076] Batch [0299]/[3760]	Speed: 69.782398 samples/sec	 accuracy=68.671875	 loss=1.282575	 lr=0.001000
Epoch[076] Batch [0349]/[3760]	Speed: 69.346879 samples/sec	 accuracy=68.593750	 loss=1.285045	 lr=0.001000
Epoch[076] Batch [0399]/[3760]	Speed: 70.925068 samples/sec	 accuracy=68.453125	 loss=1.287162	 lr=0.001000
Epoch[076] Batch [0449]/[3760]	Speed: 68.886695 samples/sec	 accuracy=68.291667	 loss=1.288239	 lr=0.001000
Epoch[076] Batch [0499]/[3760]	Speed: 70.304959 samples/sec	 accuracy=68.218750	 loss=1.289551	 lr=0.001000
Epoch[076] Batch [0549]/[3760]	Speed: 67.999519 samples/sec	 accuracy=68.284091	 loss=1.289770	 lr=0.001000
Epoch[076] Batch [0599]/[3760]	Speed: 70.539517 samples/sec	 accuracy=68.276042	 loss=1.290186	 lr=0.001000
Epoch[076] Batch [0649]/[3760]	Speed: 68.622916 samples/sec	 accuracy=68.459135	 loss=1.284686	 lr=0.001000
Epoch[076] Batch [0699]/[3760]	Speed: 70.496340 samples/sec	 accuracy=68.430804	 loss=1.284440	 lr=0.001000
Epoch[076] Batch [0749]/[3760]	Speed: 70.769790 samples/sec	 accuracy=68.412500	 loss=1.284953	 lr=0.001000
Epoch[076] Batch [0799]/[3760]	Speed: 68.948518 samples/sec	 accuracy=68.427734	 loss=1.285028	 lr=0.001000
Epoch[076] Batch [0849]/[3760]	Speed: 69.997183 samples/sec	 accuracy=68.400735	 loss=1.286033	 lr=0.001000
Epoch[076] Batch [0899]/[3760]	Speed: 68.589397 samples/sec	 accuracy=68.380208	 loss=1.285256	 lr=0.001000
Epoch[076] Batch [0949]/[3760]	Speed: 69.751496 samples/sec	 accuracy=68.435855	 loss=1.283461	 lr=0.001000
Epoch[076] Batch [0999]/[3760]	Speed: 71.276035 samples/sec	 accuracy=68.450000	 loss=1.283951	 lr=0.001000
Epoch[076] Batch [1049]/[3760]	Speed: 68.722389 samples/sec	 accuracy=68.389881	 loss=1.286735	 lr=0.001000
Epoch[076] Batch [1099]/[3760]	Speed: 69.387982 samples/sec	 accuracy=68.382102	 loss=1.286475	 lr=0.001000
Epoch[076] Batch [1149]/[3760]	Speed: 69.705236 samples/sec	 accuracy=68.432065	 loss=1.284337	 lr=0.001000
Epoch[076] Batch [1199]/[3760]	Speed: 70.626543 samples/sec	 accuracy=68.425781	 loss=1.284755	 lr=0.001000
Epoch[076] Batch [1249]/[3760]	Speed: 70.796785 samples/sec	 accuracy=68.372500	 loss=1.286057	 lr=0.001000
Epoch[076] Batch [1299]/[3760]	Speed: 68.533163 samples/sec	 accuracy=68.320913	 loss=1.287781	 lr=0.001000
Epoch[076] Batch [1349]/[3760]	Speed: 65.610889 samples/sec	 accuracy=68.363426	 loss=1.286309	 lr=0.001000
Epoch[076] Batch [1399]/[3760]	Speed: 72.104275 samples/sec	 accuracy=68.322545	 loss=1.287873	 lr=0.001000
Epoch[076] Batch [1449]/[3760]	Speed: 70.575038 samples/sec	 accuracy=68.329741	 loss=1.287880	 lr=0.001000
Epoch[076] Batch [1499]/[3760]	Speed: 68.891573 samples/sec	 accuracy=68.372917	 loss=1.286194	 lr=0.001000
Epoch[076] Batch [1549]/[3760]	Speed: 70.645678 samples/sec	 accuracy=68.375000	 loss=1.286429	 lr=0.001000
Epoch[076] Batch [1599]/[3760]	Speed: 68.886558 samples/sec	 accuracy=68.386719	 loss=1.286692	 lr=0.001000
Epoch[076] Batch [1649]/[3760]	Speed: 71.152258 samples/sec	 accuracy=68.392045	 loss=1.287066	 lr=0.001000
Epoch[076] Batch [1699]/[3760]	Speed: 69.058424 samples/sec	 accuracy=68.393382	 loss=1.287038	 lr=0.001000
Epoch[076] Batch [1749]/[3760]	Speed: 68.238835 samples/sec	 accuracy=68.344643	 loss=1.288571	 lr=0.001000
Epoch[076] Batch [1799]/[3760]	Speed: 70.802138 samples/sec	 accuracy=68.357639	 loss=1.288738	 lr=0.001000
Epoch[076] Batch [1849]/[3760]	Speed: 69.178100 samples/sec	 accuracy=68.362331	 loss=1.287752	 lr=0.001000
Epoch[076] Batch [1899]/[3760]	Speed: 69.138586 samples/sec	 accuracy=68.347039	 loss=1.287769	 lr=0.001000
Epoch[076] Batch [1949]/[3760]	Speed: 69.765595 samples/sec	 accuracy=68.325321	 loss=1.287314	 lr=0.001000
Epoch[076] Batch [1999]/[3760]	Speed: 70.216035 samples/sec	 accuracy=68.371875	 loss=1.287008	 lr=0.001000
Epoch[076] Batch [2049]/[3760]	Speed: 69.739316 samples/sec	 accuracy=68.358994	 loss=1.287584	 lr=0.001000
Epoch[076] Batch [2099]/[3760]	Speed: 69.342601 samples/sec	 accuracy=68.354911	 loss=1.287883	 lr=0.001000
Epoch[076] Batch [2149]/[3760]	Speed: 70.162573 samples/sec	 accuracy=68.359012	 loss=1.288376	 lr=0.001000
Epoch[076] Batch [2199]/[3760]	Speed: 69.730361 samples/sec	 accuracy=68.357244	 loss=1.288081	 lr=0.001000
Epoch[076] Batch [2249]/[3760]	Speed: 70.615519 samples/sec	 accuracy=68.355556	 loss=1.288023	 lr=0.001000
Epoch[076] Batch [2299]/[3760]	Speed: 69.448229 samples/sec	 accuracy=68.324728	 loss=1.288905	 lr=0.001000
Epoch[076] Batch [2349]/[3760]	Speed: 69.241833 samples/sec	 accuracy=68.297872	 loss=1.288843	 lr=0.001000
Epoch[076] Batch [2399]/[3760]	Speed: 69.553330 samples/sec	 accuracy=68.291016	 loss=1.288739	 lr=0.001000
Epoch[076] Batch [2449]/[3760]	Speed: 69.399898 samples/sec	 accuracy=68.272321	 loss=1.289319	 lr=0.001000
Epoch[076] Batch [2499]/[3760]	Speed: 70.753499 samples/sec	 accuracy=68.253125	 loss=1.289946	 lr=0.001000
Epoch[076] Batch [2549]/[3760]	Speed: 69.227509 samples/sec	 accuracy=68.240809	 loss=1.290752	 lr=0.001000
Epoch[076] Batch [2599]/[3760]	Speed: 69.181458 samples/sec	 accuracy=68.228365	 loss=1.290988	 lr=0.001000
Epoch[076] Batch [2649]/[3760]	Speed: 71.723716 samples/sec	 accuracy=68.208137	 loss=1.291103	 lr=0.001000
Epoch[076] Batch [2699]/[3760]	Speed: 69.729661 samples/sec	 accuracy=68.212384	 loss=1.290563	 lr=0.001000
Epoch[076] Batch [2749]/[3760]	Speed: 69.376710 samples/sec	 accuracy=68.219318	 loss=1.290189	 lr=0.001000
Epoch[076] Batch [2799]/[3760]	Speed: 69.173998 samples/sec	 accuracy=68.187500	 loss=1.291413	 lr=0.001000
Epoch[076] Batch [2849]/[3760]	Speed: 71.082648 samples/sec	 accuracy=68.171601	 loss=1.291959	 lr=0.001000
Epoch[076] Batch [2899]/[3760]	Speed: 68.521410 samples/sec	 accuracy=68.179418	 loss=1.291828	 lr=0.001000
Epoch[076] Batch [2949]/[3760]	Speed: 70.793674 samples/sec	 accuracy=68.181674	 loss=1.291401	 lr=0.001000
Epoch[076] Batch [2999]/[3760]	Speed: 69.811796 samples/sec	 accuracy=68.155729	 loss=1.292782	 lr=0.001000
Epoch[076] Batch [3049]/[3760]	Speed: 69.618700 samples/sec	 accuracy=68.150615	 loss=1.293251	 lr=0.001000
Epoch[076] Batch [3099]/[3760]	Speed: 69.182954 samples/sec	 accuracy=68.136593	 loss=1.293163	 lr=0.001000
Epoch[076] Batch [3149]/[3760]	Speed: 70.551587 samples/sec	 accuracy=68.121032	 loss=1.293158	 lr=0.001000
Epoch[076] Batch [3199]/[3760]	Speed: 70.105648 samples/sec	 accuracy=68.129395	 loss=1.292898	 lr=0.001000
Epoch[076] Batch [3249]/[3760]	Speed: 69.165967 samples/sec	 accuracy=68.135577	 loss=1.292886	 lr=0.001000
Epoch[076] Batch [3299]/[3760]	Speed: 69.116440 samples/sec	 accuracy=68.144886	 loss=1.292093	 lr=0.001000
Epoch[076] Batch [3349]/[3760]	Speed: 70.598481 samples/sec	 accuracy=68.149720	 loss=1.291881	 lr=0.001000
Epoch[076] Batch [3399]/[3760]	Speed: 69.971994 samples/sec	 accuracy=68.127757	 loss=1.292889	 lr=0.001000
Epoch[076] Batch [3449]/[3760]	Speed: 67.941655 samples/sec	 accuracy=68.134964	 loss=1.292996	 lr=0.001000
Epoch[076] Batch [3499]/[3760]	Speed: 71.219806 samples/sec	 accuracy=68.170982	 loss=1.292184	 lr=0.001000
Epoch[076] Batch [3549]/[3760]	Speed: 69.443947 samples/sec	 accuracy=68.166373	 loss=1.292243	 lr=0.001000
Epoch[076] Batch [3599]/[3760]	Speed: 69.608427 samples/sec	 accuracy=68.162326	 loss=1.293125	 lr=0.001000
Epoch[076] Batch [3649]/[3760]	Speed: 68.332310 samples/sec	 accuracy=68.158390	 loss=1.293345	 lr=0.001000
Epoch[076] Batch [3699]/[3760]	Speed: 71.368534 samples/sec	 accuracy=68.152872	 loss=1.293723	 lr=0.001000
Epoch[076] Batch [3749]/[3760]	Speed: 75.870365 samples/sec	 accuracy=68.155833	 loss=1.293944	 lr=0.001000
Batch [0049]/[0303]: acc-top1=62.218750 acc-top5=83.656250
Batch [0099]/[0303]: acc-top1=62.046875 acc-top5=83.515625
Batch [0149]/[0303]: acc-top1=62.958333 acc-top5=83.864583
Batch [0199]/[0303]: acc-top1=63.445312 acc-top5=84.320312
Batch [0249]/[0303]: acc-top1=63.643750 acc-top5=84.593750
Batch [0299]/[0303]: acc-top1=63.406250 acc-top5=84.526042
[Epoch 076] training: accuracy=68.150349	 loss=1.294015
[Epoch 076] speed: 69 samples/sec	time cost: 3754.415083
[Epoch 076] validation: acc-top1=63.371493 acc-top5=84.540017 loss=1.605317
Epoch[077] Batch [0049]/[3760]	Speed: 48.349116 samples/sec	 accuracy=70.031250	 loss=1.187950	 lr=0.001000
Epoch[077] Batch [0099]/[3760]	Speed: 68.606653 samples/sec	 accuracy=69.343750	 loss=1.231368	 lr=0.001000
Epoch[077] Batch [0149]/[3760]	Speed: 68.367034 samples/sec	 accuracy=68.875000	 loss=1.250620	 lr=0.001000
Epoch[077] Batch [0199]/[3760]	Speed: 70.267909 samples/sec	 accuracy=68.750000	 loss=1.263162	 lr=0.001000
Epoch[077] Batch [0249]/[3760]	Speed: 68.212146 samples/sec	 accuracy=68.543750	 loss=1.268039	 lr=0.001000
Epoch[077] Batch [0299]/[3760]	Speed: 71.689204 samples/sec	 accuracy=68.468750	 loss=1.268510	 lr=0.001000
Epoch[077] Batch [0349]/[3760]	Speed: 70.179023 samples/sec	 accuracy=68.397321	 loss=1.274068	 lr=0.001000
Epoch[077] Batch [0399]/[3760]	Speed: 68.887786 samples/sec	 accuracy=68.656250	 loss=1.266021	 lr=0.001000
Epoch[077] Batch [0449]/[3760]	Speed: 70.008440 samples/sec	 accuracy=68.440972	 loss=1.275423	 lr=0.001000
Epoch[077] Batch [0499]/[3760]	Speed: 69.432165 samples/sec	 accuracy=68.528125	 loss=1.273764	 lr=0.001000
Epoch[077] Batch [0549]/[3760]	Speed: 68.127106 samples/sec	 accuracy=68.386364	 loss=1.279948	 lr=0.001000
Epoch[077] Batch [0599]/[3760]	Speed: 70.568218 samples/sec	 accuracy=68.291667	 loss=1.281907	 lr=0.001000
Epoch[077] Batch [0649]/[3760]	Speed: 66.313268 samples/sec	 accuracy=68.295673	 loss=1.280491	 lr=0.001000
Epoch[077] Batch [0699]/[3760]	Speed: 73.834819 samples/sec	 accuracy=68.265625	 loss=1.282324	 lr=0.001000
Epoch[077] Batch [0749]/[3760]	Speed: 68.419121 samples/sec	 accuracy=68.227083	 loss=1.283496	 lr=0.001000
Epoch[077] Batch [0799]/[3760]	Speed: 70.199665 samples/sec	 accuracy=68.291016	 loss=1.283187	 lr=0.001000
Epoch[077] Batch [0849]/[3760]	Speed: 69.877913 samples/sec	 accuracy=68.354779	 loss=1.282331	 lr=0.001000
Epoch[077] Batch [0899]/[3760]	Speed: 69.489767 samples/sec	 accuracy=68.319444	 loss=1.283400	 lr=0.001000
Epoch[077] Batch [0949]/[3760]	Speed: 70.612473 samples/sec	 accuracy=68.315789	 loss=1.283885	 lr=0.001000
Epoch[077] Batch [0999]/[3760]	Speed: 69.205525 samples/sec	 accuracy=68.278125	 loss=1.284800	 lr=0.001000
Epoch[077] Batch [1049]/[3760]	Speed: 70.724624 samples/sec	 accuracy=68.226190	 loss=1.286769	 lr=0.001000
Epoch[077] Batch [1099]/[3760]	Speed: 68.414525 samples/sec	 accuracy=68.221591	 loss=1.285963	 lr=0.001000
Epoch[077] Batch [1149]/[3760]	Speed: 70.295846 samples/sec	 accuracy=68.305707	 loss=1.284279	 lr=0.001000
Epoch[077] Batch [1199]/[3760]	Speed: 70.247924 samples/sec	 accuracy=68.285156	 loss=1.283393	 lr=0.001000
Epoch[077] Batch [1249]/[3760]	Speed: 69.680937 samples/sec	 accuracy=68.232500	 loss=1.284512	 lr=0.001000
Epoch[077] Batch [1299]/[3760]	Speed: 70.337254 samples/sec	 accuracy=68.217548	 loss=1.285610	 lr=0.001000
Epoch[077] Batch [1349]/[3760]	Speed: 68.848493 samples/sec	 accuracy=68.163194	 loss=1.287761	 lr=0.001000
Epoch[077] Batch [1399]/[3760]	Speed: 69.737038 samples/sec	 accuracy=68.175223	 loss=1.288701	 lr=0.001000
Epoch[077] Batch [1449]/[3760]	Speed: 70.423870 samples/sec	 accuracy=68.191810	 loss=1.289426	 lr=0.001000
Epoch[077] Batch [1499]/[3760]	Speed: 70.819604 samples/sec	 accuracy=68.193750	 loss=1.288817	 lr=0.001000
Epoch[077] Batch [1549]/[3760]	Speed: 68.556360 samples/sec	 accuracy=68.190524	 loss=1.289179	 lr=0.001000
Epoch[077] Batch [1599]/[3760]	Speed: 69.381618 samples/sec	 accuracy=68.179688	 loss=1.288432	 lr=0.001000
Epoch[077] Batch [1649]/[3760]	Speed: 71.790331 samples/sec	 accuracy=68.178030	 loss=1.288829	 lr=0.001000
Epoch[077] Batch [1699]/[3760]	Speed: 68.831361 samples/sec	 accuracy=68.180147	 loss=1.288469	 lr=0.001000
Epoch[077] Batch [1749]/[3760]	Speed: 71.432525 samples/sec	 accuracy=68.204464	 loss=1.288428	 lr=0.001000
Epoch[077] Batch [1799]/[3760]	Speed: 68.689535 samples/sec	 accuracy=68.185764	 loss=1.288533	 lr=0.001000
Epoch[077] Batch [1849]/[3760]	Speed: 69.258496 samples/sec	 accuracy=68.148649	 loss=1.290616	 lr=0.001000
Epoch[077] Batch [1899]/[3760]	Speed: 70.208697 samples/sec	 accuracy=68.187500	 loss=1.290013	 lr=0.001000
Epoch[077] Batch [1949]/[3760]	Speed: 70.236328 samples/sec	 accuracy=68.195513	 loss=1.289581	 lr=0.001000
Epoch[077] Batch [1999]/[3760]	Speed: 69.134489 samples/sec	 accuracy=68.214844	 loss=1.288715	 lr=0.001000
Epoch[077] Batch [2049]/[3760]	Speed: 70.239371 samples/sec	 accuracy=68.218750	 loss=1.288855	 lr=0.001000
Epoch[077] Batch [2099]/[3760]	Speed: 67.413418 samples/sec	 accuracy=68.258185	 loss=1.288183	 lr=0.001000
Epoch[077] Batch [2149]/[3760]	Speed: 71.486115 samples/sec	 accuracy=68.259448	 loss=1.288058	 lr=0.001000
Epoch[077] Batch [2199]/[3760]	Speed: 69.888799 samples/sec	 accuracy=68.265625	 loss=1.287870	 lr=0.001000
Epoch[077] Batch [2249]/[3760]	Speed: 66.070412 samples/sec	 accuracy=68.281250	 loss=1.287049	 lr=0.001000
Epoch[077] Batch [2299]/[3760]	Speed: 73.471718 samples/sec	 accuracy=68.273777	 loss=1.287158	 lr=0.001000
Epoch[077] Batch [2349]/[3760]	Speed: 69.237299 samples/sec	 accuracy=68.265293	 loss=1.287447	 lr=0.001000
Epoch[077] Batch [2399]/[3760]	Speed: 70.414358 samples/sec	 accuracy=68.253255	 loss=1.287846	 lr=0.001000
Epoch[077] Batch [2449]/[3760]	Speed: 68.418992 samples/sec	 accuracy=68.214923	 loss=1.289645	 lr=0.001000
Epoch[077] Batch [2499]/[3760]	Speed: 70.196878 samples/sec	 accuracy=68.181250	 loss=1.290762	 lr=0.001000
Epoch[077] Batch [2549]/[3760]	Speed: 69.946869 samples/sec	 accuracy=68.181373	 loss=1.290242	 lr=0.001000
Epoch[077] Batch [2599]/[3760]	Speed: 69.156625 samples/sec	 accuracy=68.184495	 loss=1.290473	 lr=0.001000
Epoch[077] Batch [2649]/[3760]	Speed: 70.366041 samples/sec	 accuracy=68.166863	 loss=1.291128	 lr=0.001000
Epoch[077] Batch [2699]/[3760]	Speed: 69.290298 samples/sec	 accuracy=68.166088	 loss=1.291054	 lr=0.001000
Epoch[077] Batch [2749]/[3760]	Speed: 68.942523 samples/sec	 accuracy=68.161932	 loss=1.291474	 lr=0.001000
Epoch[077] Batch [2799]/[3760]	Speed: 68.708647 samples/sec	 accuracy=68.137835	 loss=1.291862	 lr=0.001000
Epoch[077] Batch [2849]/[3760]	Speed: 71.362983 samples/sec	 accuracy=68.114583	 loss=1.292697	 lr=0.001000
Epoch[077] Batch [2899]/[3760]	Speed: 69.974316 samples/sec	 accuracy=68.077586	 loss=1.294203	 lr=0.001000
Epoch[077] Batch [2949]/[3760]	Speed: 70.142521 samples/sec	 accuracy=68.101695	 loss=1.293772	 lr=0.001000
Epoch[077] Batch [2999]/[3760]	Speed: 69.412188 samples/sec	 accuracy=68.095313	 loss=1.293940	 lr=0.001000
Epoch[077] Batch [3049]/[3760]	Speed: 70.460775 samples/sec	 accuracy=68.096824	 loss=1.293269	 lr=0.001000
Epoch[077] Batch [3099]/[3760]	Speed: 67.884060 samples/sec	 accuracy=68.104335	 loss=1.292616	 lr=0.001000
Epoch[077] Batch [3149]/[3760]	Speed: 70.826159 samples/sec	 accuracy=68.104167	 loss=1.293080	 lr=0.001000
Epoch[077] Batch [3199]/[3760]	Speed: 69.506031 samples/sec	 accuracy=68.084961	 loss=1.294053	 lr=0.001000
Epoch[077] Batch [3249]/[3760]	Speed: 69.096645 samples/sec	 accuracy=68.096154	 loss=1.294114	 lr=0.001000
Epoch[077] Batch [3299]/[3760]	Speed: 69.834391 samples/sec	 accuracy=68.088068	 loss=1.294411	 lr=0.001000
Epoch[077] Batch [3349]/[3760]	Speed: 68.303592 samples/sec	 accuracy=68.086287	 loss=1.294818	 lr=0.001000
Epoch[077] Batch [3399]/[3760]	Speed: 70.812290 samples/sec	 accuracy=68.094669	 loss=1.294115	 lr=0.001000
Epoch[077] Batch [3449]/[3760]	Speed: 70.756220 samples/sec	 accuracy=68.100091	 loss=1.294494	 lr=0.001000
Epoch[077] Batch [3499]/[3760]	Speed: 69.280877 samples/sec	 accuracy=68.079464	 loss=1.295283	 lr=0.001000
Epoch[077] Batch [3549]/[3760]	Speed: 70.486143 samples/sec	 accuracy=68.091109	 loss=1.295127	 lr=0.001000
Epoch[077] Batch [3599]/[3760]	Speed: 70.521898 samples/sec	 accuracy=68.101562	 loss=1.294634	 lr=0.001000
Epoch[077] Batch [3649]/[3760]	Speed: 69.531990 samples/sec	 accuracy=68.098459	 loss=1.294896	 lr=0.001000
Epoch[077] Batch [3699]/[3760]	Speed: 70.408138 samples/sec	 accuracy=68.106419	 loss=1.294594	 lr=0.001000
Epoch[077] Batch [3749]/[3760]	Speed: 76.901098 samples/sec	 accuracy=68.103333	 loss=1.294439	 lr=0.001000
Batch [0049]/[0303]: acc-top1=62.000000 acc-top5=83.187500
Batch [0099]/[0303]: acc-top1=61.875000 acc-top5=83.156250
Batch [0149]/[0303]: acc-top1=63.114583 acc-top5=83.677083
Batch [0199]/[0303]: acc-top1=63.554687 acc-top5=84.109375
Batch [0249]/[0303]: acc-top1=63.718750 acc-top5=84.387500
Batch [0299]/[0303]: acc-top1=63.593750 acc-top5=84.328125
[Epoch 077] training: accuracy=68.104222	 loss=1.294376
[Epoch 077] speed: 69 samples/sec	time cost: 3749.462037
[Epoch 077] validation: acc-top1=63.593234 acc-top5=84.333746 loss=1.604057
Epoch[078] Batch [0049]/[3759]	Speed: 49.140034 samples/sec	 accuracy=68.750000	 loss=1.248820	 lr=0.001000
Epoch[078] Batch [0099]/[3759]	Speed: 68.000082 samples/sec	 accuracy=69.281250	 loss=1.230867	 lr=0.001000
Epoch[078] Batch [0149]/[3759]	Speed: 69.537901 samples/sec	 accuracy=68.656250	 loss=1.255107	 lr=0.001000
Epoch[078] Batch [0199]/[3759]	Speed: 70.323344 samples/sec	 accuracy=68.710938	 loss=1.253672	 lr=0.001000
Epoch[078] Batch [0249]/[3759]	Speed: 68.902901 samples/sec	 accuracy=68.693750	 loss=1.255802	 lr=0.001000
Epoch[078] Batch [0299]/[3759]	Speed: 70.539953 samples/sec	 accuracy=68.697917	 loss=1.260659	 lr=0.001000
Epoch[078] Batch [0349]/[3759]	Speed: 69.279399 samples/sec	 accuracy=68.799107	 loss=1.255881	 lr=0.001000
Epoch[078] Batch [0399]/[3759]	Speed: 69.681635 samples/sec	 accuracy=68.812500	 loss=1.251988	 lr=0.001000
Epoch[078] Batch [0449]/[3759]	Speed: 69.971497 samples/sec	 accuracy=68.777778	 loss=1.255718	 lr=0.001000
Epoch[078] Batch [0499]/[3759]	Speed: 69.484422 samples/sec	 accuracy=68.559375	 loss=1.263254	 lr=0.001000
Epoch[078] Batch [0549]/[3759]	Speed: 69.926311 samples/sec	 accuracy=68.585227	 loss=1.265175	 lr=0.001000
Epoch[078] Batch [0599]/[3759]	Speed: 70.261218 samples/sec	 accuracy=68.645833	 loss=1.264395	 lr=0.001000
Epoch[078] Batch [0649]/[3759]	Speed: 69.018163 samples/sec	 accuracy=68.653846	 loss=1.261591	 lr=0.001000
Epoch[078] Batch [0699]/[3759]	Speed: 70.239858 samples/sec	 accuracy=68.533482	 loss=1.265183	 lr=0.001000
Epoch[078] Batch [0749]/[3759]	Speed: 70.660674 samples/sec	 accuracy=68.575000	 loss=1.265167	 lr=0.001000
Epoch[078] Batch [0799]/[3759]	Speed: 70.066232 samples/sec	 accuracy=68.521484	 loss=1.266883	 lr=0.001000
Epoch[078] Batch [0849]/[3759]	Speed: 69.351447 samples/sec	 accuracy=68.435662	 loss=1.268407	 lr=0.001000
Epoch[078] Batch [0899]/[3759]	Speed: 69.735410 samples/sec	 accuracy=68.447917	 loss=1.268347	 lr=0.001000
Epoch[078] Batch [0949]/[3759]	Speed: 70.416203 samples/sec	 accuracy=68.467105	 loss=1.267591	 lr=0.001000
Epoch[078] Batch [0999]/[3759]	Speed: 69.788919 samples/sec	 accuracy=68.514062	 loss=1.267397	 lr=0.001000
Epoch[078] Batch [1049]/[3759]	Speed: 69.278595 samples/sec	 accuracy=68.537202	 loss=1.266818	 lr=0.001000
Epoch[078] Batch [1099]/[3759]	Speed: 70.140796 samples/sec	 accuracy=68.532670	 loss=1.268420	 lr=0.001000
Epoch[078] Batch [1149]/[3759]	Speed: 68.958944 samples/sec	 accuracy=68.483696	 loss=1.269997	 lr=0.001000
Epoch[078] Batch [1199]/[3759]	Speed: 71.041886 samples/sec	 accuracy=68.467448	 loss=1.270796	 lr=0.001000
Epoch[078] Batch [1249]/[3759]	Speed: 69.989788 samples/sec	 accuracy=68.448750	 loss=1.272895	 lr=0.001000
Epoch[078] Batch [1299]/[3759]	Speed: 70.962410 samples/sec	 accuracy=68.460337	 loss=1.272759	 lr=0.001000
Epoch[078] Batch [1349]/[3759]	Speed: 67.725496 samples/sec	 accuracy=68.483796	 loss=1.271225	 lr=0.001000
Epoch[078] Batch [1399]/[3759]	Speed: 70.644157 samples/sec	 accuracy=68.543527	 loss=1.269392	 lr=0.001000
Epoch[078] Batch [1449]/[3759]	Speed: 70.396243 samples/sec	 accuracy=68.511853	 loss=1.269655	 lr=0.001000
Epoch[078] Batch [1499]/[3759]	Speed: 70.694714 samples/sec	 accuracy=68.470833	 loss=1.271028	 lr=0.001000
Epoch[078] Batch [1549]/[3759]	Speed: 69.427301 samples/sec	 accuracy=68.408266	 loss=1.273025	 lr=0.001000
Epoch[078] Batch [1599]/[3759]	Speed: 69.011422 samples/sec	 accuracy=68.399414	 loss=1.273675	 lr=0.001000
Epoch[078] Batch [1649]/[3759]	Speed: 70.605259 samples/sec	 accuracy=68.413826	 loss=1.274500	 lr=0.001000
Epoch[078] Batch [1699]/[3759]	Speed: 69.908988 samples/sec	 accuracy=68.420037	 loss=1.274447	 lr=0.001000
Epoch[078] Batch [1749]/[3759]	Speed: 68.549825 samples/sec	 accuracy=68.432143	 loss=1.274520	 lr=0.001000
Epoch[078] Batch [1799]/[3759]	Speed: 70.959095 samples/sec	 accuracy=68.421875	 loss=1.275099	 lr=0.001000
Epoch[078] Batch [1849]/[3759]	Speed: 69.432119 samples/sec	 accuracy=68.409628	 loss=1.275933	 lr=0.001000
Epoch[078] Batch [1899]/[3759]	Speed: 70.254253 samples/sec	 accuracy=68.361020	 loss=1.277556	 lr=0.001000
Epoch[078] Batch [1949]/[3759]	Speed: 68.368996 samples/sec	 accuracy=68.366186	 loss=1.277842	 lr=0.001000
Epoch[078] Batch [1999]/[3759]	Speed: 72.236312 samples/sec	 accuracy=68.395313	 loss=1.277544	 lr=0.001000
Epoch[078] Batch [2049]/[3759]	Speed: 69.939926 samples/sec	 accuracy=68.418445	 loss=1.276984	 lr=0.001000
Epoch[078] Batch [2099]/[3759]	Speed: 68.803250 samples/sec	 accuracy=68.424851	 loss=1.277225	 lr=0.001000
Epoch[078] Batch [2149]/[3759]	Speed: 67.768821 samples/sec	 accuracy=68.404070	 loss=1.278012	 lr=0.001000
Epoch[078] Batch [2199]/[3759]	Speed: 71.912049 samples/sec	 accuracy=68.422585	 loss=1.277766	 lr=0.001000
Epoch[078] Batch [2249]/[3759]	Speed: 69.411486 samples/sec	 accuracy=68.425000	 loss=1.278104	 lr=0.001000
Epoch[078] Batch [2299]/[3759]	Speed: 68.570132 samples/sec	 accuracy=68.424592	 loss=1.277957	 lr=0.001000
Epoch[078] Batch [2349]/[3759]	Speed: 69.742852 samples/sec	 accuracy=68.408910	 loss=1.278376	 lr=0.001000
Epoch[078] Batch [2399]/[3759]	Speed: 70.371185 samples/sec	 accuracy=68.421224	 loss=1.278040	 lr=0.001000
Epoch[078] Batch [2449]/[3759]	Speed: 70.538146 samples/sec	 accuracy=68.428571	 loss=1.277366	 lr=0.001000
Epoch[078] Batch [2499]/[3759]	Speed: 70.019710 samples/sec	 accuracy=68.416875	 loss=1.278160	 lr=0.001000
Epoch[078] Batch [2549]/[3759]	Speed: 70.013306 samples/sec	 accuracy=68.436275	 loss=1.277419	 lr=0.001000
Epoch[078] Batch [2599]/[3759]	Speed: 69.517531 samples/sec	 accuracy=68.441106	 loss=1.277658	 lr=0.001000
Epoch[078] Batch [2649]/[3759]	Speed: 67.458776 samples/sec	 accuracy=68.428066	 loss=1.277788	 lr=0.001000
Epoch[078] Batch [2699]/[3759]	Speed: 70.597669 samples/sec	 accuracy=68.394097	 loss=1.279175	 lr=0.001000
Epoch[078] Batch [2749]/[3759]	Speed: 70.677185 samples/sec	 accuracy=68.398864	 loss=1.279445	 lr=0.001000
Epoch[078] Batch [2799]/[3759]	Speed: 68.669879 samples/sec	 accuracy=68.382812	 loss=1.279989	 lr=0.001000
Epoch[078] Batch [2849]/[3759]	Speed: 69.908376 samples/sec	 accuracy=68.348684	 loss=1.281001	 lr=0.001000
Epoch[078] Batch [2899]/[3759]	Speed: 68.701671 samples/sec	 accuracy=68.360991	 loss=1.281064	 lr=0.001000
Epoch[078] Batch [2949]/[3759]	Speed: 71.163247 samples/sec	 accuracy=68.331568	 loss=1.281922	 lr=0.001000
Epoch[078] Batch [2999]/[3759]	Speed: 71.115569 samples/sec	 accuracy=68.317188	 loss=1.282752	 lr=0.001000
Epoch[078] Batch [3049]/[3759]	Speed: 67.911946 samples/sec	 accuracy=68.324795	 loss=1.282789	 lr=0.001000
Epoch[078] Batch [3099]/[3759]	Speed: 70.381277 samples/sec	 accuracy=68.326613	 loss=1.283247	 lr=0.001000
Epoch[078] Batch [3149]/[3759]	Speed: 69.413851 samples/sec	 accuracy=68.342262	 loss=1.282484	 lr=0.001000
Epoch[078] Batch [3199]/[3759]	Speed: 70.016817 samples/sec	 accuracy=68.346191	 loss=1.281979	 lr=0.001000
Epoch[078] Batch [3249]/[3759]	Speed: 68.827387 samples/sec	 accuracy=68.345673	 loss=1.282072	 lr=0.001000
Epoch[078] Batch [3299]/[3759]	Speed: 70.968908 samples/sec	 accuracy=68.350379	 loss=1.282432	 lr=0.001000
Epoch[078] Batch [3349]/[3759]	Speed: 69.496461 samples/sec	 accuracy=68.329291	 loss=1.283271	 lr=0.001000
Epoch[078] Batch [3399]/[3759]	Speed: 67.742979 samples/sec	 accuracy=68.304228	 loss=1.283940	 lr=0.001000
Epoch[078] Batch [3449]/[3759]	Speed: 70.401592 samples/sec	 accuracy=68.303895	 loss=1.284129	 lr=0.001000
Epoch[078] Batch [3499]/[3759]	Speed: 70.145579 samples/sec	 accuracy=68.292411	 loss=1.284847	 lr=0.001000
Epoch[078] Batch [3549]/[3759]	Speed: 69.402293 samples/sec	 accuracy=68.275088	 loss=1.285459	 lr=0.001000
Epoch[078] Batch [3599]/[3759]	Speed: 68.776057 samples/sec	 accuracy=68.268663	 loss=1.286183	 lr=0.001000
Epoch[078] Batch [3649]/[3759]	Speed: 71.067308 samples/sec	 accuracy=68.246575	 loss=1.287016	 lr=0.001000
Epoch[078] Batch [3699]/[3759]	Speed: 69.226231 samples/sec	 accuracy=68.258868	 loss=1.286744	 lr=0.001000
Epoch[078] Batch [3749]/[3759]	Speed: 77.243668 samples/sec	 accuracy=68.247083	 loss=1.287180	 lr=0.001000
Batch [0049]/[0303]: acc-top1=61.656250 acc-top5=83.312500
Batch [0099]/[0303]: acc-top1=61.390625 acc-top5=83.703125
Batch [0149]/[0303]: acc-top1=62.479167 acc-top5=84.083333
Batch [0199]/[0303]: acc-top1=62.984375 acc-top5=84.468750
Batch [0249]/[0303]: acc-top1=63.287500 acc-top5=84.668750
Batch [0299]/[0303]: acc-top1=63.119792 acc-top5=84.546875
[Epoch 078] training: accuracy=68.250366	 loss=1.287086
[Epoch 078] speed: 69 samples/sec	time cost: 3747.882806
[Epoch 078] validation: acc-top1=63.108498 acc-top5=84.560644 loss=1.598771
Epoch[079] Batch [0049]/[3760]	Speed: 48.624452 samples/sec	 accuracy=68.843750	 loss=1.256779	 lr=0.001000
Epoch[079] Batch [0099]/[3760]	Speed: 69.561025 samples/sec	 accuracy=68.343750	 loss=1.272876	 lr=0.001000
Epoch[079] Batch [0149]/[3760]	Speed: 69.971693 samples/sec	 accuracy=68.333333	 loss=1.280803	 lr=0.001000
Epoch[079] Batch [0199]/[3760]	Speed: 69.212614 samples/sec	 accuracy=68.156250	 loss=1.286530	 lr=0.001000
Epoch[079] Batch [0249]/[3760]	Speed: 69.210340 samples/sec	 accuracy=68.537500	 loss=1.268631	 lr=0.001000
Epoch[079] Batch [0299]/[3760]	Speed: 68.621840 samples/sec	 accuracy=68.411458	 loss=1.271189	 lr=0.001000
Epoch[079] Batch [0349]/[3760]	Speed: 70.829249 samples/sec	 accuracy=68.433036	 loss=1.268198	 lr=0.001000
Epoch[079] Batch [0399]/[3760]	Speed: 70.309776 samples/sec	 accuracy=68.488281	 loss=1.267990	 lr=0.001000
Epoch[079] Batch [0449]/[3760]	Speed: 68.941706 samples/sec	 accuracy=68.350694	 loss=1.271133	 lr=0.001000
Epoch[079] Batch [0499]/[3760]	Speed: 70.246355 samples/sec	 accuracy=68.421875	 loss=1.268360	 lr=0.001000
Epoch[079] Batch [0549]/[3760]	Speed: 68.913623 samples/sec	 accuracy=68.485795	 loss=1.268951	 lr=0.001000
Epoch[079] Batch [0599]/[3760]	Speed: 69.162098 samples/sec	 accuracy=68.445312	 loss=1.273722	 lr=0.001000
Epoch[079] Batch [0649]/[3760]	Speed: 68.873137 samples/sec	 accuracy=68.437500	 loss=1.273358	 lr=0.001000
Epoch[079] Batch [0699]/[3760]	Speed: 70.044211 samples/sec	 accuracy=68.457589	 loss=1.273654	 lr=0.001000
Epoch[079] Batch [0749]/[3760]	Speed: 70.059134 samples/sec	 accuracy=68.404167	 loss=1.277071	 lr=0.001000
Epoch[079] Batch [0799]/[3760]	Speed: 68.470455 samples/sec	 accuracy=68.302734	 loss=1.281335	 lr=0.001000
Epoch[079] Batch [0849]/[3760]	Speed: 71.252383 samples/sec	 accuracy=68.343750	 loss=1.282186	 lr=0.001000
Epoch[079] Batch [0899]/[3760]	Speed: 67.735233 samples/sec	 accuracy=68.197917	 loss=1.286993	 lr=0.001000
Epoch[079] Batch [0949]/[3760]	Speed: 70.803662 samples/sec	 accuracy=68.167763	 loss=1.287747	 lr=0.001000
Epoch[079] Batch [0999]/[3760]	Speed: 69.670922 samples/sec	 accuracy=68.106250	 loss=1.289255	 lr=0.001000
Epoch[079] Batch [1049]/[3760]	Speed: 70.534351 samples/sec	 accuracy=68.142857	 loss=1.287518	 lr=0.001000
Epoch[079] Batch [1099]/[3760]	Speed: 69.660900 samples/sec	 accuracy=68.190341	 loss=1.287143	 lr=0.001000
Epoch[079] Batch [1149]/[3760]	Speed: 69.738689 samples/sec	 accuracy=68.188859	 loss=1.287115	 lr=0.001000
Epoch[079] Batch [1199]/[3760]	Speed: 69.964059 samples/sec	 accuracy=68.250000	 loss=1.285352	 lr=0.001000
Epoch[079] Batch [1249]/[3760]	Speed: 68.469990 samples/sec	 accuracy=68.225000	 loss=1.286084	 lr=0.001000
Epoch[079] Batch [1299]/[3760]	Speed: 68.978093 samples/sec	 accuracy=68.276442	 loss=1.284086	 lr=0.001000
Epoch[079] Batch [1349]/[3760]	Speed: 70.480097 samples/sec	 accuracy=68.298611	 loss=1.284924	 lr=0.001000
Epoch[079] Batch [1399]/[3760]	Speed: 68.492991 samples/sec	 accuracy=68.279018	 loss=1.285330	 lr=0.001000
Epoch[079] Batch [1449]/[3760]	Speed: 71.471272 samples/sec	 accuracy=68.287716	 loss=1.285638	 lr=0.001000
Epoch[079] Batch [1499]/[3760]	Speed: 69.244737 samples/sec	 accuracy=68.268750	 loss=1.287808	 lr=0.001000
Epoch[079] Batch [1549]/[3760]	Speed: 70.305365 samples/sec	 accuracy=68.240927	 loss=1.288345	 lr=0.001000
Epoch[079] Batch [1599]/[3760]	Speed: 65.383087 samples/sec	 accuracy=68.208008	 loss=1.289294	 lr=0.001000
Epoch[079] Batch [1649]/[3760]	Speed: 71.888952 samples/sec	 accuracy=68.208333	 loss=1.289110	 lr=0.001000
Epoch[079] Batch [1699]/[3760]	Speed: 70.102056 samples/sec	 accuracy=68.188419	 loss=1.289458	 lr=0.001000
Epoch[079] Batch [1749]/[3760]	Speed: 70.401748 samples/sec	 accuracy=68.178571	 loss=1.289293	 lr=0.001000
Epoch[079] Batch [1799]/[3760]	Speed: 69.538764 samples/sec	 accuracy=68.173611	 loss=1.290009	 lr=0.001000
Epoch[079] Batch [1849]/[3760]	Speed: 69.117494 samples/sec	 accuracy=68.152027	 loss=1.289905	 lr=0.001000
Epoch[079] Batch [1899]/[3760]	Speed: 70.853911 samples/sec	 accuracy=68.166118	 loss=1.289923	 lr=0.001000
Epoch[079] Batch [1949]/[3760]	Speed: 69.667970 samples/sec	 accuracy=68.145833	 loss=1.290698	 lr=0.001000
Epoch[079] Batch [1999]/[3760]	Speed: 69.990131 samples/sec	 accuracy=68.138281	 loss=1.290817	 lr=0.001000
Epoch[079] Batch [2049]/[3760]	Speed: 70.413147 samples/sec	 accuracy=68.173018	 loss=1.289556	 lr=0.001000
Epoch[079] Batch [2099]/[3760]	Speed: 68.610280 samples/sec	 accuracy=68.212054	 loss=1.288465	 lr=0.001000
Epoch[079] Batch [2149]/[3760]	Speed: 70.543776 samples/sec	 accuracy=68.205669	 loss=1.288835	 lr=0.001000
Epoch[079] Batch [2199]/[3760]	Speed: 69.765039 samples/sec	 accuracy=68.242188	 loss=1.287586	 lr=0.001000
Epoch[079] Batch [2249]/[3760]	Speed: 70.665148 samples/sec	 accuracy=68.253472	 loss=1.286824	 lr=0.001000
Epoch[079] Batch [2299]/[3760]	Speed: 67.190545 samples/sec	 accuracy=68.222147	 loss=1.287798	 lr=0.001000
Epoch[079] Batch [2349]/[3760]	Speed: 71.512616 samples/sec	 accuracy=68.234043	 loss=1.287135	 lr=0.001000
Epoch[079] Batch [2399]/[3760]	Speed: 70.381690 samples/sec	 accuracy=68.227865	 loss=1.287585	 lr=0.001000
Epoch[079] Batch [2449]/[3760]	Speed: 69.985111 samples/sec	 accuracy=68.241071	 loss=1.288048	 lr=0.001000
Epoch[079] Batch [2499]/[3760]	Speed: 69.256371 samples/sec	 accuracy=68.230000	 loss=1.288284	 lr=0.001000
Epoch[079] Batch [2549]/[3760]	Speed: 69.718870 samples/sec	 accuracy=68.241422	 loss=1.288288	 lr=0.001000
Epoch[079] Batch [2599]/[3760]	Speed: 69.634471 samples/sec	 accuracy=68.248798	 loss=1.288859	 lr=0.001000
Epoch[079] Batch [2649]/[3760]	Speed: 69.085576 samples/sec	 accuracy=68.261203	 loss=1.288301	 lr=0.001000
Epoch[079] Batch [2699]/[3760]	Speed: 70.642952 samples/sec	 accuracy=68.270833	 loss=1.287732	 lr=0.001000
Epoch[079] Batch [2749]/[3760]	Speed: 67.250298 samples/sec	 accuracy=68.257386	 loss=1.288788	 lr=0.001000
Epoch[079] Batch [2799]/[3760]	Speed: 71.861843 samples/sec	 accuracy=68.238839	 loss=1.289724	 lr=0.001000
Epoch[079] Batch [2849]/[3760]	Speed: 69.152611 samples/sec	 accuracy=68.255482	 loss=1.288744	 lr=0.001000
Epoch[079] Batch [2899]/[3760]	Speed: 70.851911 samples/sec	 accuracy=68.251078	 loss=1.288496	 lr=0.001000
Epoch[079] Batch [2949]/[3760]	Speed: 69.552176 samples/sec	 accuracy=68.247881	 loss=1.288814	 lr=0.001000
Epoch[079] Batch [2999]/[3760]	Speed: 67.861255 samples/sec	 accuracy=68.242188	 loss=1.289112	 lr=0.001000
Epoch[079] Batch [3049]/[3760]	Speed: 71.725626 samples/sec	 accuracy=68.256148	 loss=1.289041	 lr=0.001000
Epoch[079] Batch [3099]/[3760]	Speed: 67.446526 samples/sec	 accuracy=68.263609	 loss=1.288863	 lr=0.001000
Epoch[079] Batch [3149]/[3760]	Speed: 68.547019 samples/sec	 accuracy=68.271825	 loss=1.288905	 lr=0.001000
Epoch[079] Batch [3199]/[3760]	Speed: 72.054419 samples/sec	 accuracy=68.282227	 loss=1.288509	 lr=0.001000
Epoch[079] Batch [3249]/[3760]	Speed: 71.060540 samples/sec	 accuracy=68.286058	 loss=1.288345	 lr=0.001000
Epoch[079] Batch [3299]/[3760]	Speed: 70.183002 samples/sec	 accuracy=68.282197	 loss=1.287869	 lr=0.001000
Epoch[079] Batch [3349]/[3760]	Speed: 69.146993 samples/sec	 accuracy=68.274254	 loss=1.288324	 lr=0.001000
Epoch[079] Batch [3399]/[3760]	Speed: 69.991349 samples/sec	 accuracy=68.264706	 loss=1.288358	 lr=0.001000
Epoch[079] Batch [3449]/[3760]	Speed: 69.479902 samples/sec	 accuracy=68.285326	 loss=1.287638	 lr=0.001000
Epoch[079] Batch [3499]/[3760]	Speed: 69.782837 samples/sec	 accuracy=68.273214	 loss=1.287812	 lr=0.001000
Epoch[079] Batch [3549]/[3760]	Speed: 70.809834 samples/sec	 accuracy=68.262324	 loss=1.288131	 lr=0.001000
Epoch[079] Batch [3599]/[3760]	Speed: 70.645649 samples/sec	 accuracy=68.250868	 loss=1.288475	 lr=0.001000
Epoch[079] Batch [3649]/[3760]	Speed: 68.729217 samples/sec	 accuracy=68.251284	 loss=1.288319	 lr=0.001000
Epoch[079] Batch [3699]/[3760]	Speed: 69.051375 samples/sec	 accuracy=68.255912	 loss=1.288457	 lr=0.001000
Epoch[079] Batch [3749]/[3760]	Speed: 77.310806 samples/sec	 accuracy=68.255417	 loss=1.288437	 lr=0.000100
Batch [0049]/[0303]: acc-top1=62.343750 acc-top5=84.156250
Batch [0099]/[0303]: acc-top1=62.265625 acc-top5=84.515625
Batch [0149]/[0303]: acc-top1=63.187500 acc-top5=84.583333
Batch [0199]/[0303]: acc-top1=63.726562 acc-top5=84.914062
Batch [0249]/[0303]: acc-top1=63.856250 acc-top5=85.087500
Batch [0299]/[0303]: acc-top1=63.692708 acc-top5=84.854167
[Epoch 079] training: accuracy=68.256732	 loss=1.288498
[Epoch 079] speed: 69 samples/sec	time cost: 3753.424043
[Epoch 079] validation: acc-top1=63.670586 acc-top5=84.859736 loss=1.592256
Epoch[080] Batch [0049]/[3760]	Speed: 49.100707 samples/sec	 accuracy=67.625000	 loss=1.290182	 lr=0.000100
Epoch[080] Batch [0099]/[3760]	Speed: 68.492756 samples/sec	 accuracy=68.187500	 loss=1.277923	 lr=0.000100
Epoch[080] Batch [0149]/[3760]	Speed: 67.882710 samples/sec	 accuracy=68.541667	 loss=1.276186	 lr=0.000100
Epoch[080] Batch [0199]/[3760]	Speed: 71.100544 samples/sec	 accuracy=68.867188	 loss=1.256672	 lr=0.000100
Epoch[080] Batch [0249]/[3760]	Speed: 69.579079 samples/sec	 accuracy=69.106250	 loss=1.252365	 lr=0.000100
Epoch[080] Batch [0299]/[3760]	Speed: 71.100456 samples/sec	 accuracy=68.890625	 loss=1.254826	 lr=0.000100
Epoch[080] Batch [0349]/[3760]	Speed: 69.619898 samples/sec	 accuracy=68.910714	 loss=1.252501	 lr=0.000100
Epoch[080] Batch [0399]/[3760]	Speed: 69.850979 samples/sec	 accuracy=68.902344	 loss=1.250958	 lr=0.000100
Epoch[080] Batch [0449]/[3760]	Speed: 70.209185 samples/sec	 accuracy=68.861111	 loss=1.252038	 lr=0.000100
Epoch[080] Batch [0499]/[3760]	Speed: 69.269918 samples/sec	 accuracy=68.984375	 loss=1.249064	 lr=0.000100
Epoch[080] Batch [0549]/[3760]	Speed: 70.059211 samples/sec	 accuracy=69.017045	 loss=1.249882	 lr=0.000100
Epoch[080] Batch [0599]/[3760]	Speed: 67.034944 samples/sec	 accuracy=69.026042	 loss=1.251046	 lr=0.000100
Epoch[080] Batch [0649]/[3760]	Speed: 72.076585 samples/sec	 accuracy=69.038462	 loss=1.250707	 lr=0.000100
Epoch[080] Batch [0699]/[3760]	Speed: 70.631486 samples/sec	 accuracy=69.024554	 loss=1.250712	 lr=0.000100
Epoch[080] Batch [0749]/[3760]	Speed: 68.414245 samples/sec	 accuracy=69.033333	 loss=1.250623	 lr=0.000100
Epoch[080] Batch [0799]/[3760]	Speed: 70.898375 samples/sec	 accuracy=69.056641	 loss=1.249274	 lr=0.000100
Epoch[080] Batch [0849]/[3760]	Speed: 69.672536 samples/sec	 accuracy=69.080882	 loss=1.247564	 lr=0.000100
Epoch[080] Batch [0899]/[3760]	Speed: 70.977029 samples/sec	 accuracy=69.055556	 loss=1.248945	 lr=0.000100
Epoch[080] Batch [0949]/[3760]	Speed: 69.098259 samples/sec	 accuracy=69.023026	 loss=1.250196	 lr=0.000100
Epoch[080] Batch [0999]/[3760]	Speed: 69.448685 samples/sec	 accuracy=69.117188	 loss=1.247280	 lr=0.000100
Epoch[080] Batch [1049]/[3760]	Speed: 70.364844 samples/sec	 accuracy=69.084821	 loss=1.248386	 lr=0.000100
Epoch[080] Batch [1099]/[3760]	Speed: 68.910531 samples/sec	 accuracy=69.126420	 loss=1.247583	 lr=0.000100
Epoch[080] Batch [1149]/[3760]	Speed: 70.659839 samples/sec	 accuracy=69.146739	 loss=1.247174	 lr=0.000100
Epoch[080] Batch [1199]/[3760]	Speed: 68.597812 samples/sec	 accuracy=69.149740	 loss=1.246060	 lr=0.000100
Epoch[080] Batch [1249]/[3760]	Speed: 70.227026 samples/sec	 accuracy=69.183750	 loss=1.244920	 lr=0.000100
Epoch[080] Batch [1299]/[3760]	Speed: 69.212169 samples/sec	 accuracy=69.175481	 loss=1.243174	 lr=0.000100
Epoch[080] Batch [1349]/[3760]	Speed: 69.807074 samples/sec	 accuracy=69.245370	 loss=1.240324	 lr=0.000100
Epoch[080] Batch [1399]/[3760]	Speed: 67.613694 samples/sec	 accuracy=69.293527	 loss=1.240603	 lr=0.000100
Epoch[080] Batch [1449]/[3760]	Speed: 72.911156 samples/sec	 accuracy=69.304957	 loss=1.241502	 lr=0.000100
Epoch[080] Batch [1499]/[3760]	Speed: 68.541480 samples/sec	 accuracy=69.284375	 loss=1.242055	 lr=0.000100
Epoch[080] Batch [1549]/[3760]	Speed: 69.671594 samples/sec	 accuracy=69.330645	 loss=1.240836	 lr=0.000100
Epoch[080] Batch [1599]/[3760]	Speed: 69.467966 samples/sec	 accuracy=69.316406	 loss=1.241800	 lr=0.000100
Epoch[080] Batch [1649]/[3760]	Speed: 72.430614 samples/sec	 accuracy=69.285038	 loss=1.243341	 lr=0.000100
Epoch[080] Batch [1699]/[3760]	Speed: 68.935422 samples/sec	 accuracy=69.268382	 loss=1.242738	 lr=0.000100
Epoch[080] Batch [1749]/[3760]	Speed: 68.944403 samples/sec	 accuracy=69.262500	 loss=1.242636	 lr=0.000100
Epoch[080] Batch [1799]/[3760]	Speed: 71.086021 samples/sec	 accuracy=69.310764	 loss=1.241149	 lr=0.000100
Epoch[080] Batch [1849]/[3760]	Speed: 69.697370 samples/sec	 accuracy=69.311655	 loss=1.240491	 lr=0.000100
Epoch[080] Batch [1899]/[3760]	Speed: 69.507879 samples/sec	 accuracy=69.331414	 loss=1.239912	 lr=0.000100
Epoch[080] Batch [1949]/[3760]	Speed: 70.403338 samples/sec	 accuracy=69.362981	 loss=1.238568	 lr=0.000100
Epoch[080] Batch [1999]/[3760]	Speed: 69.104676 samples/sec	 accuracy=69.374219	 loss=1.238962	 lr=0.000100
Epoch[080] Batch [2049]/[3760]	Speed: 69.807224 samples/sec	 accuracy=69.371951	 loss=1.239261	 lr=0.000100
Epoch[080] Batch [2099]/[3760]	Speed: 70.620000 samples/sec	 accuracy=69.372768	 loss=1.240081	 lr=0.000100
Epoch[080] Batch [2149]/[3760]	Speed: 70.632555 samples/sec	 accuracy=69.385901	 loss=1.239278	 lr=0.000100
Epoch[080] Batch [2199]/[3760]	Speed: 68.860974 samples/sec	 accuracy=69.406250	 loss=1.238397	 lr=0.000100
Epoch[080] Batch [2249]/[3760]	Speed: 70.820989 samples/sec	 accuracy=69.440278	 loss=1.237213	 lr=0.000100
Epoch[080] Batch [2299]/[3760]	Speed: 69.563277 samples/sec	 accuracy=69.461277	 loss=1.236321	 lr=0.000100
Epoch[080] Batch [2349]/[3760]	Speed: 69.652822 samples/sec	 accuracy=69.470080	 loss=1.235413	 lr=0.000100
Epoch[080] Batch [2399]/[3760]	Speed: 70.732112 samples/sec	 accuracy=69.501302	 loss=1.234753	 lr=0.000100
Epoch[080] Batch [2449]/[3760]	Speed: 68.595287 samples/sec	 accuracy=69.478316	 loss=1.235081	 lr=0.000100
Epoch[080] Batch [2499]/[3760]	Speed: 70.238461 samples/sec	 accuracy=69.493125	 loss=1.234769	 lr=0.000100
Epoch[080] Batch [2549]/[3760]	Speed: 69.393861 samples/sec	 accuracy=69.473652	 loss=1.235163	 lr=0.000100
Epoch[080] Batch [2599]/[3760]	Speed: 71.001999 samples/sec	 accuracy=69.475962	 loss=1.235053	 lr=0.000100
Epoch[080] Batch [2649]/[3760]	Speed: 68.538605 samples/sec	 accuracy=69.469929	 loss=1.235481	 lr=0.000100
Epoch[080] Batch [2699]/[3760]	Speed: 70.238891 samples/sec	 accuracy=69.473958	 loss=1.235167	 lr=0.000100
Epoch[080] Batch [2749]/[3760]	Speed: 70.553294 samples/sec	 accuracy=69.481250	 loss=1.235089	 lr=0.000100
Epoch[080] Batch [2799]/[3760]	Speed: 67.716640 samples/sec	 accuracy=69.493862	 loss=1.235393	 lr=0.000100
Epoch[080] Batch [2849]/[3760]	Speed: 71.293266 samples/sec	 accuracy=69.472588	 loss=1.236580	 lr=0.000100
Epoch[080] Batch [2899]/[3760]	Speed: 69.416852 samples/sec	 accuracy=69.472522	 loss=1.236644	 lr=0.000100
Epoch[080] Batch [2949]/[3760]	Speed: 69.655032 samples/sec	 accuracy=69.472458	 loss=1.236816	 lr=0.000100
Epoch[080] Batch [2999]/[3760]	Speed: 70.095782 samples/sec	 accuracy=69.458333	 loss=1.237069	 lr=0.000100
Epoch[080] Batch [3049]/[3760]	Speed: 70.134692 samples/sec	 accuracy=69.474385	 loss=1.237145	 lr=0.000100
Epoch[080] Batch [3099]/[3760]	Speed: 67.824202 samples/sec	 accuracy=69.465726	 loss=1.237709	 lr=0.000100
Epoch[080] Batch [3149]/[3760]	Speed: 71.147764 samples/sec	 accuracy=69.469742	 loss=1.237412	 lr=0.000100
Epoch[080] Batch [3199]/[3760]	Speed: 69.624311 samples/sec	 accuracy=69.458984	 loss=1.237308	 lr=0.000100
Epoch[080] Batch [3249]/[3760]	Speed: 70.151345 samples/sec	 accuracy=69.453846	 loss=1.237062	 lr=0.000100
Epoch[080] Batch [3299]/[3760]	Speed: 70.056756 samples/sec	 accuracy=69.453125	 loss=1.236910	 lr=0.000100
Epoch[080] Batch [3349]/[3760]	Speed: 68.261442 samples/sec	 accuracy=69.449627	 loss=1.237230	 lr=0.000100
Epoch[080] Batch [3399]/[3760]	Speed: 69.844135 samples/sec	 accuracy=69.460478	 loss=1.236448	 lr=0.000100
Epoch[080] Batch [3449]/[3760]	Speed: 71.087767 samples/sec	 accuracy=69.437500	 loss=1.237505	 lr=0.000100
Epoch[080] Batch [3499]/[3760]	Speed: 69.780268 samples/sec	 accuracy=69.447321	 loss=1.237093	 lr=0.000100
Epoch[080] Batch [3549]/[3760]	Speed: 69.856410 samples/sec	 accuracy=69.445863	 loss=1.236773	 lr=0.000100
Epoch[080] Batch [3599]/[3760]	Speed: 70.516116 samples/sec	 accuracy=69.444010	 loss=1.236714	 lr=0.000100
Epoch[080] Batch [3649]/[3760]	Speed: 69.281769 samples/sec	 accuracy=69.440925	 loss=1.236920	 lr=0.000100
Epoch[080] Batch [3699]/[3760]	Speed: 69.190398 samples/sec	 accuracy=69.456503	 loss=1.236493	 lr=0.000100
Epoch[080] Batch [3749]/[3760]	Speed: 78.023107 samples/sec	 accuracy=69.462500	 loss=1.236111	 lr=0.000100
Batch [0049]/[0303]: acc-top1=62.562500 acc-top5=84.343750
Batch [0099]/[0303]: acc-top1=62.375000 acc-top5=84.390625
Batch [0149]/[0303]: acc-top1=63.437500 acc-top5=84.770833
Batch [0199]/[0303]: acc-top1=64.000000 acc-top5=85.070312
Batch [0249]/[0303]: acc-top1=64.318750 acc-top5=85.225000
Batch [0299]/[0303]: acc-top1=64.114583 acc-top5=85.031250
[Epoch 080] training: accuracy=69.470163	 loss=1.236002
[Epoch 080] speed: 69 samples/sec	time cost: 3744.252791
[Epoch 080] validation: acc-top1=64.088284 acc-top5=85.045380 loss=1.572592
Epoch[081] Batch [0049]/[3759]	Speed: 47.657623 samples/sec	 accuracy=69.125000	 loss=1.249328	 lr=0.000100
Epoch[081] Batch [0099]/[3759]	Speed: 70.925706 samples/sec	 accuracy=69.203125	 loss=1.233521	 lr=0.000100
Epoch[081] Batch [0149]/[3759]	Speed: 70.134236 samples/sec	 accuracy=69.208333	 loss=1.230737	 lr=0.000100
Epoch[081] Batch [0199]/[3759]	Speed: 69.911165 samples/sec	 accuracy=69.578125	 loss=1.228205	 lr=0.000100
Epoch[081] Batch [0249]/[3759]	Speed: 69.584695 samples/sec	 accuracy=69.687500	 loss=1.226401	 lr=0.000100
Epoch[081] Batch [0299]/[3759]	Speed: 70.535604 samples/sec	 accuracy=69.703125	 loss=1.228211	 lr=0.000100
Epoch[081] Batch [0349]/[3759]	Speed: 67.693374 samples/sec	 accuracy=69.816964	 loss=1.228433	 lr=0.000100
Epoch[081] Batch [0399]/[3759]	Speed: 70.730840 samples/sec	 accuracy=69.902344	 loss=1.226681	 lr=0.000100
Epoch[081] Batch [0449]/[3759]	Speed: 71.010142 samples/sec	 accuracy=69.944444	 loss=1.222411	 lr=0.000100
Epoch[081] Batch [0499]/[3759]	Speed: 69.175090 samples/sec	 accuracy=69.950000	 loss=1.221395	 lr=0.000100
Epoch[081] Batch [0549]/[3759]	Speed: 70.400319 samples/sec	 accuracy=69.926136	 loss=1.220822	 lr=0.000100
Epoch[081] Batch [0599]/[3759]	Speed: 69.926665 samples/sec	 accuracy=69.927083	 loss=1.220319	 lr=0.000100
Epoch[081] Batch [0649]/[3759]	Speed: 70.039594 samples/sec	 accuracy=69.915865	 loss=1.221449	 lr=0.000100
Epoch[081] Batch [0699]/[3759]	Speed: 69.954553 samples/sec	 accuracy=69.823661	 loss=1.221052	 lr=0.000100
Epoch[081] Batch [0749]/[3759]	Speed: 68.517948 samples/sec	 accuracy=69.783333	 loss=1.222400	 lr=0.000100
Epoch[081] Batch [0799]/[3759]	Speed: 69.636834 samples/sec	 accuracy=69.716797	 loss=1.225141	 lr=0.000100
Epoch[081] Batch [0849]/[3759]	Speed: 70.257162 samples/sec	 accuracy=69.761029	 loss=1.222077	 lr=0.000100
Epoch[081] Batch [0899]/[3759]	Speed: 70.923501 samples/sec	 accuracy=69.732639	 loss=1.222365	 lr=0.000100
Epoch[081] Batch [0949]/[3759]	Speed: 67.642027 samples/sec	 accuracy=69.746711	 loss=1.220979	 lr=0.000100
Epoch[081] Batch [0999]/[3759]	Speed: 70.873542 samples/sec	 accuracy=69.753125	 loss=1.221727	 lr=0.000100
Epoch[081] Batch [1049]/[3759]	Speed: 70.028677 samples/sec	 accuracy=69.748512	 loss=1.221927	 lr=0.000100
Epoch[081] Batch [1099]/[3759]	Speed: 69.499656 samples/sec	 accuracy=69.720170	 loss=1.222351	 lr=0.000100
Epoch[081] Batch [1149]/[3759]	Speed: 69.161388 samples/sec	 accuracy=69.748641	 loss=1.221579	 lr=0.000100
Epoch[081] Batch [1199]/[3759]	Speed: 70.494409 samples/sec	 accuracy=69.766927	 loss=1.220591	 lr=0.000100
Epoch[081] Batch [1249]/[3759]	Speed: 70.906792 samples/sec	 accuracy=69.777500	 loss=1.221290	 lr=0.000100
Epoch[081] Batch [1299]/[3759]	Speed: 69.080068 samples/sec	 accuracy=69.764423	 loss=1.222813	 lr=0.000100
Epoch[081] Batch [1349]/[3759]	Speed: 69.965445 samples/sec	 accuracy=69.761574	 loss=1.222821	 lr=0.000100
Epoch[081] Batch [1399]/[3759]	Speed: 70.261005 samples/sec	 accuracy=69.772321	 loss=1.223316	 lr=0.000100
Epoch[081] Batch [1449]/[3759]	Speed: 69.413875 samples/sec	 accuracy=69.786638	 loss=1.223011	 lr=0.000100
Epoch[081] Batch [1499]/[3759]	Speed: 69.360406 samples/sec	 accuracy=69.828125	 loss=1.222480	 lr=0.000100
Epoch[081] Batch [1549]/[3759]	Speed: 69.744929 samples/sec	 accuracy=69.845766	 loss=1.221811	 lr=0.000100
Epoch[081] Batch [1599]/[3759]	Speed: 70.846610 samples/sec	 accuracy=69.860352	 loss=1.221617	 lr=0.000100
Epoch[081] Batch [1649]/[3759]	Speed: 67.507386 samples/sec	 accuracy=69.862689	 loss=1.222058	 lr=0.000100
Epoch[081] Batch [1699]/[3759]	Speed: 71.690415 samples/sec	 accuracy=69.852941	 loss=1.222518	 lr=0.000100
Epoch[081] Batch [1749]/[3759]	Speed: 69.753772 samples/sec	 accuracy=69.840179	 loss=1.222515	 lr=0.000100
Epoch[081] Batch [1799]/[3759]	Speed: 69.547246 samples/sec	 accuracy=69.848958	 loss=1.220611	 lr=0.000100
Epoch[081] Batch [1849]/[3759]	Speed: 69.254296 samples/sec	 accuracy=69.844595	 loss=1.220581	 lr=0.000100
Epoch[081] Batch [1899]/[3759]	Speed: 70.404293 samples/sec	 accuracy=69.878289	 loss=1.219776	 lr=0.000100
Epoch[081] Batch [1949]/[3759]	Speed: 71.217891 samples/sec	 accuracy=69.806090	 loss=1.222028	 lr=0.000100
Epoch[081] Batch [1999]/[3759]	Speed: 68.592421 samples/sec	 accuracy=69.809375	 loss=1.222160	 lr=0.000100
Epoch[081] Batch [2049]/[3759]	Speed: 70.456620 samples/sec	 accuracy=69.837652	 loss=1.221775	 lr=0.000100
Epoch[081] Batch [2099]/[3759]	Speed: 69.060136 samples/sec	 accuracy=69.836310	 loss=1.221593	 lr=0.000100
Epoch[081] Batch [2149]/[3759]	Speed: 69.799402 samples/sec	 accuracy=69.849564	 loss=1.221389	 lr=0.000100
Epoch[081] Batch [2199]/[3759]	Speed: 68.339026 samples/sec	 accuracy=69.895597	 loss=1.219568	 lr=0.000100
Epoch[081] Batch [2249]/[3759]	Speed: 70.202657 samples/sec	 accuracy=69.912500	 loss=1.219217	 lr=0.000100
Epoch[081] Batch [2299]/[3759]	Speed: 69.379140 samples/sec	 accuracy=69.926630	 loss=1.218844	 lr=0.000100
Epoch[081] Batch [2349]/[3759]	Speed: 70.119596 samples/sec	 accuracy=69.928191	 loss=1.218222	 lr=0.000100
Epoch[081] Batch [2399]/[3759]	Speed: 70.828095 samples/sec	 accuracy=69.922526	 loss=1.218536	 lr=0.000100
Epoch[081] Batch [2449]/[3759]	Speed: 68.450145 samples/sec	 accuracy=69.949617	 loss=1.217931	 lr=0.000100
Epoch[081] Batch [2499]/[3759]	Speed: 70.480269 samples/sec	 accuracy=69.954375	 loss=1.218432	 lr=0.000100
Epoch[081] Batch [2549]/[3759]	Speed: 69.086164 samples/sec	 accuracy=69.959559	 loss=1.218015	 lr=0.000100
Epoch[081] Batch [2599]/[3759]	Speed: 69.385463 samples/sec	 accuracy=69.992788	 loss=1.216680	 lr=0.000100
Epoch[081] Batch [2649]/[3759]	Speed: 69.598635 samples/sec	 accuracy=70.000000	 loss=1.216105	 lr=0.000100
Epoch[081] Batch [2699]/[3759]	Speed: 69.701537 samples/sec	 accuracy=69.990162	 loss=1.216571	 lr=0.000100
Epoch[081] Batch [2749]/[3759]	Speed: 70.488075 samples/sec	 accuracy=69.961932	 loss=1.217266	 lr=0.000100
Epoch[081] Batch [2799]/[3759]	Speed: 70.109308 samples/sec	 accuracy=69.934710	 loss=1.218259	 lr=0.000100
Epoch[081] Batch [2849]/[3759]	Speed: 69.687669 samples/sec	 accuracy=69.942982	 loss=1.217971	 lr=0.000100
Epoch[081] Batch [2899]/[3759]	Speed: 68.492437 samples/sec	 accuracy=69.947198	 loss=1.217143	 lr=0.000100
Epoch[081] Batch [2949]/[3759]	Speed: 70.472477 samples/sec	 accuracy=69.938030	 loss=1.217571	 lr=0.000100
Epoch[081] Batch [2999]/[3759]	Speed: 68.366137 samples/sec	 accuracy=69.926042	 loss=1.218054	 lr=0.000100
Epoch[081] Batch [3049]/[3759]	Speed: 70.192125 samples/sec	 accuracy=69.916496	 loss=1.218017	 lr=0.000100
Epoch[081] Batch [3099]/[3759]	Speed: 70.510549 samples/sec	 accuracy=69.900706	 loss=1.218043	 lr=0.000100
Epoch[081] Batch [3149]/[3759]	Speed: 69.413250 samples/sec	 accuracy=69.891369	 loss=1.218014	 lr=0.000100
Epoch[081] Batch [3199]/[3759]	Speed: 70.212266 samples/sec	 accuracy=69.907715	 loss=1.217184	 lr=0.000100
Epoch[081] Batch [3249]/[3759]	Speed: 68.854887 samples/sec	 accuracy=69.891346	 loss=1.217590	 lr=0.000100
Epoch[081] Batch [3299]/[3759]	Speed: 70.828289 samples/sec	 accuracy=69.890625	 loss=1.217879	 lr=0.000100
Epoch[081] Batch [3349]/[3759]	Speed: 69.774893 samples/sec	 accuracy=69.903918	 loss=1.217320	 lr=0.000100
Epoch[081] Batch [3399]/[3759]	Speed: 69.405733 samples/sec	 accuracy=69.889246	 loss=1.217798	 lr=0.000100
Epoch[081] Batch [3449]/[3759]	Speed: 69.855428 samples/sec	 accuracy=69.879076	 loss=1.217879	 lr=0.000100
Epoch[081] Batch [3499]/[3759]	Speed: 70.176809 samples/sec	 accuracy=69.877232	 loss=1.217487	 lr=0.000100
Epoch[081] Batch [3549]/[3759]	Speed: 68.944335 samples/sec	 accuracy=69.863996	 loss=1.217710	 lr=0.000100
Epoch[081] Batch [3599]/[3759]	Speed: 70.807175 samples/sec	 accuracy=69.870226	 loss=1.217656	 lr=0.000100
Epoch[081] Batch [3649]/[3759]	Speed: 67.355934 samples/sec	 accuracy=69.877140	 loss=1.217616	 lr=0.000100
Epoch[081] Batch [3699]/[3759]	Speed: 71.903331 samples/sec	 accuracy=69.868243	 loss=1.218056	 lr=0.000100
Epoch[081] Batch [3749]/[3759]	Speed: 78.165300 samples/sec	 accuracy=69.859583	 loss=1.218445	 lr=0.000100
Batch [0049]/[0303]: acc-top1=62.718750 acc-top5=84.312500
Batch [0099]/[0303]: acc-top1=62.515625 acc-top5=84.453125
Batch [0149]/[0303]: acc-top1=63.458333 acc-top5=84.812500
Batch [0199]/[0303]: acc-top1=64.023438 acc-top5=85.125000
Batch [0249]/[0303]: acc-top1=64.368750 acc-top5=85.331250
Batch [0299]/[0303]: acc-top1=64.208333 acc-top5=85.171875
[Epoch 081] training: accuracy=69.859421	 loss=1.218616
[Epoch 081] speed: 69 samples/sec	time cost: 3746.757249
[Epoch 081] validation: acc-top1=64.186262 acc-top5=85.169142 loss=1.559025
Epoch[082] Batch [0049]/[3760]	Speed: 47.990788 samples/sec	 accuracy=70.000000	 loss=1.208133	 lr=0.000100
Epoch[082] Batch [0099]/[3760]	Speed: 68.156318 samples/sec	 accuracy=70.078125	 loss=1.213913	 lr=0.000100
Epoch[082] Batch [0149]/[3760]	Speed: 71.526411 samples/sec	 accuracy=70.177083	 loss=1.205133	 lr=0.000100
Epoch[082] Batch [0199]/[3760]	Speed: 69.076698 samples/sec	 accuracy=70.164062	 loss=1.203177	 lr=0.000100
Epoch[082] Batch [0249]/[3760]	Speed: 69.437156 samples/sec	 accuracy=70.212500	 loss=1.202410	 lr=0.000100
Epoch[082] Batch [0299]/[3760]	Speed: 70.150035 samples/sec	 accuracy=70.312500	 loss=1.198680	 lr=0.000100
Epoch[082] Batch [0349]/[3760]	Speed: 68.871438 samples/sec	 accuracy=70.116071	 loss=1.206095	 lr=0.000100
Epoch[082] Batch [0399]/[3760]	Speed: 69.235956 samples/sec	 accuracy=70.128906	 loss=1.207302	 lr=0.000100
Epoch[082] Batch [0449]/[3760]	Speed: 70.551827 samples/sec	 accuracy=70.114583	 loss=1.206901	 lr=0.000100
Epoch[082] Batch [0499]/[3760]	Speed: 67.673819 samples/sec	 accuracy=70.012500	 loss=1.210181	 lr=0.000100
Epoch[082] Batch [0549]/[3760]	Speed: 71.723117 samples/sec	 accuracy=70.048295	 loss=1.212131	 lr=0.000100
Epoch[082] Batch [0599]/[3760]	Speed: 68.720395 samples/sec	 accuracy=70.078125	 loss=1.209022	 lr=0.000100
Epoch[082] Batch [0649]/[3760]	Speed: 71.432635 samples/sec	 accuracy=70.108173	 loss=1.207994	 lr=0.000100
Epoch[082] Batch [0699]/[3760]	Speed: 67.651529 samples/sec	 accuracy=70.133929	 loss=1.206890	 lr=0.000100
Epoch[082] Batch [0749]/[3760]	Speed: 71.641240 samples/sec	 accuracy=70.227083	 loss=1.204850	 lr=0.000100
Epoch[082] Batch [0799]/[3760]	Speed: 69.509505 samples/sec	 accuracy=70.171875	 loss=1.206224	 lr=0.000100
Epoch[082] Batch [0849]/[3760]	Speed: 69.983894 samples/sec	 accuracy=70.148897	 loss=1.207819	 lr=0.000100
Epoch[082] Batch [0899]/[3760]	Speed: 67.881418 samples/sec	 accuracy=70.182292	 loss=1.208116	 lr=0.000100
Epoch[082] Batch [0949]/[3760]	Speed: 70.693771 samples/sec	 accuracy=70.154605	 loss=1.210444	 lr=0.000100
Epoch[082] Batch [0999]/[3760]	Speed: 69.613822 samples/sec	 accuracy=70.104687	 loss=1.212827	 lr=0.000100
Epoch[082] Batch [1049]/[3760]	Speed: 70.024841 samples/sec	 accuracy=70.130952	 loss=1.211944	 lr=0.000100
Epoch[082] Batch [1099]/[3760]	Speed: 68.410577 samples/sec	 accuracy=70.198864	 loss=1.209095	 lr=0.000100
Epoch[082] Batch [1149]/[3760]	Speed: 68.684620 samples/sec	 accuracy=70.226902	 loss=1.208447	 lr=0.000100
Epoch[082] Batch [1199]/[3760]	Speed: 68.477206 samples/sec	 accuracy=70.248698	 loss=1.207484	 lr=0.000100
Epoch[082] Batch [1249]/[3760]	Speed: 72.026614 samples/sec	 accuracy=70.256250	 loss=1.207242	 lr=0.000100
Epoch[082] Batch [1299]/[3760]	Speed: 70.157039 samples/sec	 accuracy=70.263221	 loss=1.207908	 lr=0.000100
Epoch[082] Batch [1349]/[3760]	Speed: 69.823144 samples/sec	 accuracy=70.209491	 loss=1.209054	 lr=0.000100
Epoch[082] Batch [1399]/[3760]	Speed: 70.792433 samples/sec	 accuracy=70.205357	 loss=1.209669	 lr=0.000100
Epoch[082] Batch [1449]/[3760]	Speed: 67.474408 samples/sec	 accuracy=70.165948	 loss=1.210524	 lr=0.000100
Epoch[082] Batch [1499]/[3760]	Speed: 71.718834 samples/sec	 accuracy=70.189583	 loss=1.210355	 lr=0.000100
Epoch[082] Batch [1549]/[3760]	Speed: 71.251278 samples/sec	 accuracy=70.185484	 loss=1.209896	 lr=0.000100
Epoch[082] Batch [1599]/[3760]	Speed: 69.204888 samples/sec	 accuracy=70.211914	 loss=1.210197	 lr=0.000100
Epoch[082] Batch [1649]/[3760]	Speed: 69.798477 samples/sec	 accuracy=70.227273	 loss=1.210072	 lr=0.000100
Epoch[082] Batch [1699]/[3760]	Speed: 69.059676 samples/sec	 accuracy=70.237132	 loss=1.210234	 lr=0.000100
Epoch[082] Batch [1749]/[3760]	Speed: 70.948680 samples/sec	 accuracy=70.260714	 loss=1.208620	 lr=0.000100
Epoch[082] Batch [1799]/[3760]	Speed: 68.377239 samples/sec	 accuracy=70.282986	 loss=1.207623	 lr=0.000100
Epoch[082] Batch [1849]/[3760]	Speed: 70.006013 samples/sec	 accuracy=70.286318	 loss=1.207701	 lr=0.000100
Epoch[082] Batch [1899]/[3760]	Speed: 70.571211 samples/sec	 accuracy=70.236020	 loss=1.209030	 lr=0.000100
Epoch[082] Batch [1949]/[3760]	Speed: 68.597084 samples/sec	 accuracy=70.212340	 loss=1.210245	 lr=0.000100
Epoch[082] Batch [1999]/[3760]	Speed: 70.211221 samples/sec	 accuracy=70.240625	 loss=1.209411	 lr=0.000100
Epoch[082] Batch [2049]/[3760]	Speed: 69.947017 samples/sec	 accuracy=70.250762	 loss=1.208818	 lr=0.000100
Epoch[082] Batch [2099]/[3760]	Speed: 68.597788 samples/sec	 accuracy=70.252976	 loss=1.209489	 lr=0.000100
Epoch[082] Batch [2149]/[3760]	Speed: 70.309584 samples/sec	 accuracy=70.271076	 loss=1.208195	 lr=0.000100
Epoch[082] Batch [2199]/[3760]	Speed: 68.033231 samples/sec	 accuracy=70.269886	 loss=1.207833	 lr=0.000100
Epoch[082] Batch [2249]/[3760]	Speed: 70.577611 samples/sec	 accuracy=70.245139	 loss=1.208281	 lr=0.000100
Epoch[082] Batch [2299]/[3760]	Speed: 71.357012 samples/sec	 accuracy=70.269701	 loss=1.207600	 lr=0.000100
Epoch[082] Batch [2349]/[3760]	Speed: 68.839855 samples/sec	 accuracy=70.254654	 loss=1.208061	 lr=0.000100
Epoch[082] Batch [2399]/[3760]	Speed: 70.490426 samples/sec	 accuracy=70.261719	 loss=1.208112	 lr=0.000100
Epoch[082] Batch [2449]/[3760]	Speed: 68.554316 samples/sec	 accuracy=70.259566	 loss=1.208613	 lr=0.000100
Epoch[082] Batch [2499]/[3760]	Speed: 70.375686 samples/sec	 accuracy=70.276250	 loss=1.207575	 lr=0.000100
Epoch[082] Batch [2549]/[3760]	Speed: 70.008311 samples/sec	 accuracy=70.264706	 loss=1.207334	 lr=0.000100
Epoch[082] Batch [2599]/[3760]	Speed: 69.641242 samples/sec	 accuracy=70.256010	 loss=1.207568	 lr=0.000100
Epoch[082] Batch [2649]/[3760]	Speed: 69.737307 samples/sec	 accuracy=70.254127	 loss=1.207203	 lr=0.000100
Epoch[082] Batch [2699]/[3760]	Speed: 68.237494 samples/sec	 accuracy=70.263310	 loss=1.206960	 lr=0.000100
Epoch[082] Batch [2749]/[3760]	Speed: 70.078726 samples/sec	 accuracy=70.271591	 loss=1.207110	 lr=0.000100
Epoch[082] Batch [2799]/[3760]	Speed: 68.844231 samples/sec	 accuracy=70.275112	 loss=1.207857	 lr=0.000100
Epoch[082] Batch [2849]/[3760]	Speed: 69.863325 samples/sec	 accuracy=70.241776	 loss=1.208961	 lr=0.000100
Epoch[082] Batch [2899]/[3760]	Speed: 70.729329 samples/sec	 accuracy=70.228987	 loss=1.209242	 lr=0.000100
Epoch[082] Batch [2949]/[3760]	Speed: 68.879009 samples/sec	 accuracy=70.225106	 loss=1.208976	 lr=0.000100
Epoch[082] Batch [2999]/[3760]	Speed: 71.372449 samples/sec	 accuracy=70.227083	 loss=1.208631	 lr=0.000100
Epoch[082] Batch [3049]/[3760]	Speed: 70.086502 samples/sec	 accuracy=70.216701	 loss=1.208921	 lr=0.000100
Epoch[082] Batch [3099]/[3760]	Speed: 65.162774 samples/sec	 accuracy=70.216230	 loss=1.208847	 lr=0.000100
Epoch[082] Batch [3149]/[3760]	Speed: 73.844728 samples/sec	 accuracy=70.198413	 loss=1.209601	 lr=0.000100
Epoch[082] Batch [3199]/[3760]	Speed: 70.356889 samples/sec	 accuracy=70.180664	 loss=1.209949	 lr=0.000100
Epoch[082] Batch [3249]/[3760]	Speed: 69.570011 samples/sec	 accuracy=70.180769	 loss=1.209705	 lr=0.000100
Epoch[082] Batch [3299]/[3760]	Speed: 69.828729 samples/sec	 accuracy=70.185606	 loss=1.209228	 lr=0.000100
Epoch[082] Batch [3349]/[3760]	Speed: 68.938234 samples/sec	 accuracy=70.187966	 loss=1.208747	 lr=0.000100
Epoch[082] Batch [3399]/[3760]	Speed: 68.178431 samples/sec	 accuracy=70.159467	 loss=1.209281	 lr=0.000100
Epoch[082] Batch [3449]/[3760]	Speed: 70.561527 samples/sec	 accuracy=70.155344	 loss=1.208826	 lr=0.000100
Epoch[082] Batch [3499]/[3760]	Speed: 70.084071 samples/sec	 accuracy=70.127679	 loss=1.209862	 lr=0.000100
Epoch[082] Batch [3549]/[3760]	Speed: 70.022801 samples/sec	 accuracy=70.129401	 loss=1.209670	 lr=0.000100
Epoch[082] Batch [3599]/[3760]	Speed: 68.121600 samples/sec	 accuracy=70.128472	 loss=1.209466	 lr=0.000100
Epoch[082] Batch [3649]/[3760]	Speed: 69.704104 samples/sec	 accuracy=70.143836	 loss=1.208933	 lr=0.000100
Epoch[082] Batch [3699]/[3760]	Speed: 70.148450 samples/sec	 accuracy=70.152449	 loss=1.208730	 lr=0.000100
Epoch[082] Batch [3749]/[3760]	Speed: 77.511173 samples/sec	 accuracy=70.143750	 loss=1.209204	 lr=0.000100
Batch [0049]/[0303]: acc-top1=62.687500 acc-top5=84.093750
Batch [0099]/[0303]: acc-top1=62.609375 acc-top5=84.250000
Batch [0149]/[0303]: acc-top1=63.687500 acc-top5=84.656250
Batch [0199]/[0303]: acc-top1=64.242188 acc-top5=85.039062
Batch [0249]/[0303]: acc-top1=64.425000 acc-top5=85.250000
Batch [0299]/[0303]: acc-top1=64.286458 acc-top5=85.088542
[Epoch 082] training: accuracy=70.149186	 loss=1.208966
[Epoch 082] speed: 69 samples/sec	time cost: 3756.365784
[Epoch 082] validation: acc-top1=64.263614 acc-top5=85.091790 loss=1.559077
Epoch[083] Batch [0049]/[3760]	Speed: 48.837152 samples/sec	 accuracy=72.281250	 loss=1.132780	 lr=0.000100
Epoch[083] Batch [0099]/[3760]	Speed: 68.077196 samples/sec	 accuracy=71.156250	 loss=1.170604	 lr=0.000100
Epoch[083] Batch [0149]/[3760]	Speed: 68.734440 samples/sec	 accuracy=70.635417	 loss=1.187446	 lr=0.000100
Epoch[083] Batch [0199]/[3760]	Speed: 68.565037 samples/sec	 accuracy=70.343750	 loss=1.190941	 lr=0.000100
Epoch[083] Batch [0249]/[3760]	Speed: 70.098713 samples/sec	 accuracy=70.612500	 loss=1.182062	 lr=0.000100
Epoch[083] Batch [0299]/[3760]	Speed: 67.878334 samples/sec	 accuracy=70.380208	 loss=1.187916	 lr=0.000100
Epoch[083] Batch [0349]/[3760]	Speed: 71.182191 samples/sec	 accuracy=70.312500	 loss=1.186928	 lr=0.000100
Epoch[083] Batch [0399]/[3760]	Speed: 69.531843 samples/sec	 accuracy=70.285156	 loss=1.190977	 lr=0.000100
Epoch[083] Batch [0449]/[3760]	Speed: 71.212483 samples/sec	 accuracy=70.190972	 loss=1.197302	 lr=0.000100
Epoch[083] Batch [0499]/[3760]	Speed: 70.218905 samples/sec	 accuracy=70.040625	 loss=1.204302	 lr=0.000100
Epoch[083] Batch [0549]/[3760]	Speed: 69.636829 samples/sec	 accuracy=70.159091	 loss=1.199583	 lr=0.000100
Epoch[083] Batch [0599]/[3760]	Speed: 68.047949 samples/sec	 accuracy=70.049479	 loss=1.201964	 lr=0.000100
Epoch[083] Batch [0649]/[3760]	Speed: 68.707811 samples/sec	 accuracy=70.079327	 loss=1.200524	 lr=0.000100
Epoch[083] Batch [0699]/[3760]	Speed: 71.160885 samples/sec	 accuracy=70.000000	 loss=1.202739	 lr=0.000100
Epoch[083] Batch [0749]/[3760]	Speed: 70.131456 samples/sec	 accuracy=70.072917	 loss=1.202163	 lr=0.000100
Epoch[083] Batch [0799]/[3760]	Speed: 67.192694 samples/sec	 accuracy=70.082031	 loss=1.203186	 lr=0.000100
Epoch[083] Batch [0849]/[3760]	Speed: 71.853656 samples/sec	 accuracy=70.044118	 loss=1.204476	 lr=0.000100
Epoch[083] Batch [0899]/[3760]	Speed: 68.879495 samples/sec	 accuracy=69.996528	 loss=1.206019	 lr=0.000100
Epoch[083] Batch [0949]/[3760]	Speed: 71.919748 samples/sec	 accuracy=70.044408	 loss=1.204573	 lr=0.000100
Epoch[083] Batch [0999]/[3760]	Speed: 68.157582 samples/sec	 accuracy=70.017188	 loss=1.205811	 lr=0.000100
Epoch[083] Batch [1049]/[3760]	Speed: 68.503266 samples/sec	 accuracy=69.985119	 loss=1.205823	 lr=0.000100
Epoch[083] Batch [1099]/[3760]	Speed: 71.279049 samples/sec	 accuracy=69.951705	 loss=1.207228	 lr=0.000100
Epoch[083] Batch [1149]/[3760]	Speed: 68.473072 samples/sec	 accuracy=69.994565	 loss=1.206598	 lr=0.000100
Epoch[083] Batch [1199]/[3760]	Speed: 70.730389 samples/sec	 accuracy=70.007812	 loss=1.208029	 lr=0.000100
Epoch[083] Batch [1249]/[3760]	Speed: 67.931865 samples/sec	 accuracy=70.021250	 loss=1.208052	 lr=0.000100
Epoch[083] Batch [1299]/[3760]	Speed: 71.325230 samples/sec	 accuracy=70.026442	 loss=1.206835	 lr=0.000100
Epoch[083] Batch [1349]/[3760]	Speed: 70.108838 samples/sec	 accuracy=69.979167	 loss=1.208901	 lr=0.000100
Epoch[083] Batch [1399]/[3760]	Speed: 65.918782 samples/sec	 accuracy=70.017857	 loss=1.210159	 lr=0.000100
Epoch[083] Batch [1449]/[3760]	Speed: 73.226050 samples/sec	 accuracy=70.039871	 loss=1.209731	 lr=0.000100
Epoch[083] Batch [1499]/[3760]	Speed: 69.999856 samples/sec	 accuracy=69.979167	 loss=1.212270	 lr=0.000100
Epoch[083] Batch [1549]/[3760]	Speed: 69.548274 samples/sec	 accuracy=69.996976	 loss=1.211400	 lr=0.000100
Epoch[083] Batch [1599]/[3760]	Speed: 69.139750 samples/sec	 accuracy=70.055664	 loss=1.209462	 lr=0.000100
Epoch[083] Batch [1649]/[3760]	Speed: 70.033713 samples/sec	 accuracy=70.071023	 loss=1.208545	 lr=0.000100
Epoch[083] Batch [1699]/[3760]	Speed: 69.989466 samples/sec	 accuracy=70.093750	 loss=1.208434	 lr=0.000100
Epoch[083] Batch [1749]/[3760]	Speed: 69.396684 samples/sec	 accuracy=70.110714	 loss=1.207739	 lr=0.000100
Epoch[083] Batch [1799]/[3760]	Speed: 68.827770 samples/sec	 accuracy=70.112847	 loss=1.207575	 lr=0.000100
Epoch[083] Batch [1849]/[3760]	Speed: 68.405223 samples/sec	 accuracy=70.114020	 loss=1.207024	 lr=0.000100
Epoch[083] Batch [1899]/[3760]	Speed: 71.327072 samples/sec	 accuracy=70.097862	 loss=1.207428	 lr=0.000100
Epoch[083] Batch [1949]/[3760]	Speed: 66.852918 samples/sec	 accuracy=70.147436	 loss=1.206107	 lr=0.000100
Epoch[083] Batch [1999]/[3760]	Speed: 71.358504 samples/sec	 accuracy=70.160156	 loss=1.206297	 lr=0.000100
Epoch[083] Batch [2049]/[3760]	Speed: 68.348447 samples/sec	 accuracy=70.172256	 loss=1.205495	 lr=0.000100
Epoch[083] Batch [2099]/[3760]	Speed: 69.252803 samples/sec	 accuracy=70.170387	 loss=1.205952	 lr=0.000100
Epoch[083] Batch [2149]/[3760]	Speed: 72.282144 samples/sec	 accuracy=70.222384	 loss=1.204050	 lr=0.000100
Epoch[083] Batch [2199]/[3760]	Speed: 68.428843 samples/sec	 accuracy=70.220881	 loss=1.203979	 lr=0.000100
Epoch[083] Batch [2249]/[3760]	Speed: 70.592668 samples/sec	 accuracy=70.214583	 loss=1.204824	 lr=0.000100
Epoch[083] Batch [2299]/[3760]	Speed: 69.828698 samples/sec	 accuracy=70.187500	 loss=1.205260	 lr=0.000100
Epoch[083] Batch [2349]/[3760]	Speed: 69.112968 samples/sec	 accuracy=70.199468	 loss=1.204860	 lr=0.000100
Epoch[083] Batch [2399]/[3760]	Speed: 71.067370 samples/sec	 accuracy=70.187500	 loss=1.205383	 lr=0.000100
Epoch[083] Batch [2449]/[3760]	Speed: 68.580864 samples/sec	 accuracy=70.206633	 loss=1.204880	 lr=0.000100
Epoch[083] Batch [2499]/[3760]	Speed: 70.379980 samples/sec	 accuracy=70.235000	 loss=1.204088	 lr=0.000100
Epoch[083] Batch [2549]/[3760]	Speed: 67.961602 samples/sec	 accuracy=70.221814	 loss=1.203804	 lr=0.000100
Epoch[083] Batch [2599]/[3760]	Speed: 68.989933 samples/sec	 accuracy=70.221154	 loss=1.203888	 lr=0.000100
Epoch[083] Batch [2649]/[3760]	Speed: 71.800764 samples/sec	 accuracy=70.227594	 loss=1.203456	 lr=0.000100
Epoch[083] Batch [2699]/[3760]	Speed: 69.162140 samples/sec	 accuracy=70.213542	 loss=1.203915	 lr=0.000100
Epoch[083] Batch [2749]/[3760]	Speed: 70.665282 samples/sec	 accuracy=70.206818	 loss=1.203775	 lr=0.000100
Epoch[083] Batch [2799]/[3760]	Speed: 69.790062 samples/sec	 accuracy=70.194196	 loss=1.204921	 lr=0.000100
Epoch[083] Batch [2849]/[3760]	Speed: 70.269962 samples/sec	 accuracy=70.203399	 loss=1.204429	 lr=0.000100
Epoch[083] Batch [2899]/[3760]	Speed: 68.856540 samples/sec	 accuracy=70.199892	 loss=1.204343	 lr=0.000100
Epoch[083] Batch [2949]/[3760]	Speed: 69.025888 samples/sec	 accuracy=70.221928	 loss=1.203691	 lr=0.000100
Epoch[083] Batch [2999]/[3760]	Speed: 71.729519 samples/sec	 accuracy=70.216146	 loss=1.203944	 lr=0.000100
Epoch[083] Batch [3049]/[3760]	Speed: 69.700796 samples/sec	 accuracy=70.216701	 loss=1.203982	 lr=0.000100
Epoch[083] Batch [3099]/[3760]	Speed: 69.892720 samples/sec	 accuracy=70.214718	 loss=1.204002	 lr=0.000100
Epoch[083] Batch [3149]/[3760]	Speed: 69.815797 samples/sec	 accuracy=70.205357	 loss=1.204443	 lr=0.000100
Epoch[083] Batch [3199]/[3760]	Speed: 68.306703 samples/sec	 accuracy=70.201660	 loss=1.204253	 lr=0.000100
Epoch[083] Batch [3249]/[3760]	Speed: 70.666769 samples/sec	 accuracy=70.200000	 loss=1.204055	 lr=0.000100
Epoch[083] Batch [3299]/[3760]	Speed: 69.685510 samples/sec	 accuracy=70.210227	 loss=1.203631	 lr=0.000100
Epoch[083] Batch [3349]/[3760]	Speed: 69.761106 samples/sec	 accuracy=70.219683	 loss=1.202751	 lr=0.000100
Epoch[083] Batch [3399]/[3760]	Speed: 68.039009 samples/sec	 accuracy=70.215074	 loss=1.203211	 lr=0.000100
Epoch[083] Batch [3449]/[3760]	Speed: 70.733248 samples/sec	 accuracy=70.202899	 loss=1.203956	 lr=0.000100
Epoch[083] Batch [3499]/[3760]	Speed: 68.969679 samples/sec	 accuracy=70.200000	 loss=1.204554	 lr=0.000100
Epoch[083] Batch [3549]/[3760]	Speed: 70.951794 samples/sec	 accuracy=70.199824	 loss=1.204762	 lr=0.000100
Epoch[083] Batch [3599]/[3760]	Speed: 70.329156 samples/sec	 accuracy=70.198351	 loss=1.205049	 lr=0.000100
Epoch[083] Batch [3649]/[3760]	Speed: 68.734478 samples/sec	 accuracy=70.182791	 loss=1.204799	 lr=0.000100
Epoch[083] Batch [3699]/[3760]	Speed: 70.499657 samples/sec	 accuracy=70.187078	 loss=1.204783	 lr=0.000100
Epoch[083] Batch [3749]/[3760]	Speed: 76.933961 samples/sec	 accuracy=70.182083	 loss=1.204916	 lr=0.000100
Batch [0049]/[0303]: acc-top1=62.656250 acc-top5=84.125000
Batch [0099]/[0303]: acc-top1=62.640625 acc-top5=84.187500
Batch [0149]/[0303]: acc-top1=63.812500 acc-top5=84.604167
Batch [0199]/[0303]: acc-top1=64.304688 acc-top5=84.992188
Batch [0249]/[0303]: acc-top1=64.606250 acc-top5=85.275000
Batch [0299]/[0303]: acc-top1=64.416667 acc-top5=85.109375
[Epoch 083] training: accuracy=70.185339	 loss=1.205017
[Epoch 083] speed: 69 samples/sec	time cost: 3755.288259
[Epoch 083] validation: acc-top1=64.387376 acc-top5=85.112417 loss=1.563435
Epoch[084] Batch [0049]/[3759]	Speed: 48.540663 samples/sec	 accuracy=68.625000	 loss=1.280062	 lr=0.000100
Epoch[084] Batch [0099]/[3759]	Speed: 67.077179 samples/sec	 accuracy=69.906250	 loss=1.235906	 lr=0.000100
Epoch[084] Batch [0149]/[3759]	Speed: 68.602187 samples/sec	 accuracy=69.843750	 loss=1.231580	 lr=0.000100
Epoch[084] Batch [0199]/[3759]	Speed: 70.377487 samples/sec	 accuracy=70.117188	 loss=1.216111	 lr=0.000100
Epoch[084] Batch [0249]/[3759]	Speed: 69.910680 samples/sec	 accuracy=70.168750	 loss=1.219911	 lr=0.000100
Epoch[084] Batch [0299]/[3759]	Speed: 70.374678 samples/sec	 accuracy=70.265625	 loss=1.208262	 lr=0.000100
Epoch[084] Batch [0349]/[3759]	Speed: 69.517207 samples/sec	 accuracy=70.397321	 loss=1.201624	 lr=0.000100
Epoch[084] Batch [0399]/[3759]	Speed: 70.189382 samples/sec	 accuracy=70.355469	 loss=1.202361	 lr=0.000100
Epoch[084] Batch [0449]/[3759]	Speed: 69.261005 samples/sec	 accuracy=70.496528	 loss=1.197206	 lr=0.000100
Epoch[084] Batch [0499]/[3759]	Speed: 70.491231 samples/sec	 accuracy=70.400000	 loss=1.196577	 lr=0.000100
Epoch[084] Batch [0549]/[3759]	Speed: 69.823362 samples/sec	 accuracy=70.414773	 loss=1.196432	 lr=0.000100
Epoch[084] Batch [0599]/[3759]	Speed: 68.048424 samples/sec	 accuracy=70.395833	 loss=1.197047	 lr=0.000100
Epoch[084] Batch [0649]/[3759]	Speed: 70.572488 samples/sec	 accuracy=70.466346	 loss=1.194866	 lr=0.000100
Epoch[084] Batch [0699]/[3759]	Speed: 69.699852 samples/sec	 accuracy=70.383929	 loss=1.200171	 lr=0.000100
Epoch[084] Batch [0749]/[3759]	Speed: 69.472897 samples/sec	 accuracy=70.291667	 loss=1.200447	 lr=0.000100
Epoch[084] Batch [0799]/[3759]	Speed: 69.502342 samples/sec	 accuracy=70.330078	 loss=1.199961	 lr=0.000100
Epoch[084] Batch [0849]/[3759]	Speed: 69.519027 samples/sec	 accuracy=70.279412	 loss=1.203016	 lr=0.000100
Epoch[084] Batch [0899]/[3759]	Speed: 70.933969 samples/sec	 accuracy=70.255208	 loss=1.205144	 lr=0.000100
Epoch[084] Batch [0949]/[3759]	Speed: 69.625938 samples/sec	 accuracy=70.213816	 loss=1.206146	 lr=0.000100
Epoch[084] Batch [0999]/[3759]	Speed: 70.724659 samples/sec	 accuracy=70.179688	 loss=1.207569	 lr=0.000100
Epoch[084] Batch [1049]/[3759]	Speed: 69.913277 samples/sec	 accuracy=70.244048	 loss=1.205471	 lr=0.000100
Epoch[084] Batch [1099]/[3759]	Speed: 68.230047 samples/sec	 accuracy=70.132102	 loss=1.209642	 lr=0.000100
Epoch[084] Batch [1149]/[3759]	Speed: 69.751173 samples/sec	 accuracy=70.187500	 loss=1.207688	 lr=0.000100
Epoch[084] Batch [1199]/[3759]	Speed: 70.084083 samples/sec	 accuracy=70.147135	 loss=1.207625	 lr=0.000100
Epoch[084] Batch [1249]/[3759]	Speed: 70.736327 samples/sec	 accuracy=70.170000	 loss=1.206173	 lr=0.000100
Epoch[084] Batch [1299]/[3759]	Speed: 68.357327 samples/sec	 accuracy=70.173077	 loss=1.205695	 lr=0.000100
Epoch[084] Batch [1349]/[3759]	Speed: 70.698643 samples/sec	 accuracy=70.185185	 loss=1.204192	 lr=0.000100
Epoch[084] Batch [1399]/[3759]	Speed: 69.358239 samples/sec	 accuracy=70.179688	 loss=1.205245	 lr=0.000100
Epoch[084] Batch [1449]/[3759]	Speed: 70.253862 samples/sec	 accuracy=70.198276	 loss=1.204687	 lr=0.000100
Epoch[084] Batch [1499]/[3759]	Speed: 68.141731 samples/sec	 accuracy=70.219792	 loss=1.204378	 lr=0.000100
Epoch[084] Batch [1549]/[3759]	Speed: 68.840136 samples/sec	 accuracy=70.233871	 loss=1.204309	 lr=0.000100
Epoch[084] Batch [1599]/[3759]	Speed: 70.406727 samples/sec	 accuracy=70.238281	 loss=1.204166	 lr=0.000100
Epoch[084] Batch [1649]/[3759]	Speed: 70.592148 samples/sec	 accuracy=70.259470	 loss=1.203845	 lr=0.000100
Epoch[084] Batch [1699]/[3759]	Speed: 69.010337 samples/sec	 accuracy=70.254596	 loss=1.203881	 lr=0.000100
Epoch[084] Batch [1749]/[3759]	Speed: 69.750633 samples/sec	 accuracy=70.268750	 loss=1.203177	 lr=0.000100
Epoch[084] Batch [1799]/[3759]	Speed: 70.609816 samples/sec	 accuracy=70.298611	 loss=1.201889	 lr=0.000100
Epoch[084] Batch [1849]/[3759]	Speed: 67.859730 samples/sec	 accuracy=70.337838	 loss=1.201558	 lr=0.000100
Epoch[084] Batch [1899]/[3759]	Speed: 71.326247 samples/sec	 accuracy=70.324836	 loss=1.202167	 lr=0.000100
Epoch[084] Batch [1949]/[3759]	Speed: 70.310248 samples/sec	 accuracy=70.299679	 loss=1.202726	 lr=0.000100
Epoch[084] Batch [1999]/[3759]	Speed: 68.891935 samples/sec	 accuracy=70.289844	 loss=1.202447	 lr=0.000100
Epoch[084] Batch [2049]/[3759]	Speed: 70.429903 samples/sec	 accuracy=70.297256	 loss=1.201707	 lr=0.000100
Epoch[084] Batch [2099]/[3759]	Speed: 70.723054 samples/sec	 accuracy=70.317708	 loss=1.201085	 lr=0.000100
Epoch[084] Batch [2149]/[3759]	Speed: 69.023716 samples/sec	 accuracy=70.314680	 loss=1.200504	 lr=0.000100
Epoch[084] Batch [2199]/[3759]	Speed: 68.144534 samples/sec	 accuracy=70.311790	 loss=1.201102	 lr=0.000100
Epoch[084] Batch [2249]/[3759]	Speed: 71.036146 samples/sec	 accuracy=70.322222	 loss=1.201218	 lr=0.000100
Epoch[084] Batch [2299]/[3759]	Speed: 70.063000 samples/sec	 accuracy=70.343750	 loss=1.200221	 lr=0.000100
Epoch[084] Batch [2349]/[3759]	Speed: 67.589395 samples/sec	 accuracy=70.347739	 loss=1.199939	 lr=0.000100
Epoch[084] Batch [2399]/[3759]	Speed: 71.477997 samples/sec	 accuracy=70.347005	 loss=1.200145	 lr=0.000100
Epoch[084] Batch [2449]/[3759]	Speed: 69.853482 samples/sec	 accuracy=70.332270	 loss=1.200481	 lr=0.000100
Epoch[084] Batch [2499]/[3759]	Speed: 68.875183 samples/sec	 accuracy=70.346250	 loss=1.200166	 lr=0.000100
Epoch[084] Batch [2549]/[3759]	Speed: 71.074484 samples/sec	 accuracy=70.327206	 loss=1.200962	 lr=0.000100
Epoch[084] Batch [2599]/[3759]	Speed: 69.221691 samples/sec	 accuracy=70.325721	 loss=1.200620	 lr=0.000100
Epoch[084] Batch [2649]/[3759]	Speed: 69.148685 samples/sec	 accuracy=70.327241	 loss=1.200671	 lr=0.000100
Epoch[084] Batch [2699]/[3759]	Speed: 69.545564 samples/sec	 accuracy=70.322338	 loss=1.201256	 lr=0.000100
Epoch[084] Batch [2749]/[3759]	Speed: 70.522030 samples/sec	 accuracy=70.336364	 loss=1.200210	 lr=0.000100
Epoch[084] Batch [2799]/[3759]	Speed: 69.907775 samples/sec	 accuracy=70.339286	 loss=1.200400	 lr=0.000100
Epoch[084] Batch [2849]/[3759]	Speed: 70.120717 samples/sec	 accuracy=70.343202	 loss=1.199880	 lr=0.000100
Epoch[084] Batch [2899]/[3759]	Speed: 68.320359 samples/sec	 accuracy=70.333513	 loss=1.199655	 lr=0.000100
Epoch[084] Batch [2949]/[3759]	Speed: 68.978215 samples/sec	 accuracy=70.359110	 loss=1.198954	 lr=0.000100
Epoch[084] Batch [2999]/[3759]	Speed: 71.708694 samples/sec	 accuracy=70.369792	 loss=1.198566	 lr=0.000100
Epoch[084] Batch [3049]/[3759]	Speed: 67.723442 samples/sec	 accuracy=70.353484	 loss=1.198972	 lr=0.000100
Epoch[084] Batch [3099]/[3759]	Speed: 70.349777 samples/sec	 accuracy=70.328125	 loss=1.199409	 lr=0.000100
Epoch[084] Batch [3149]/[3759]	Speed: 71.386826 samples/sec	 accuracy=70.345734	 loss=1.198754	 lr=0.000100
Epoch[084] Batch [3199]/[3759]	Speed: 68.082837 samples/sec	 accuracy=70.334473	 loss=1.199105	 lr=0.000100
Epoch[084] Batch [3249]/[3759]	Speed: 69.639043 samples/sec	 accuracy=70.324038	 loss=1.199087	 lr=0.000100
Epoch[084] Batch [3299]/[3759]	Speed: 69.037689 samples/sec	 accuracy=70.332860	 loss=1.198766	 lr=0.000100
Epoch[084] Batch [3349]/[3759]	Speed: 70.016695 samples/sec	 accuracy=70.333022	 loss=1.199596	 lr=0.000100
Epoch[084] Batch [3399]/[3759]	Speed: 71.139791 samples/sec	 accuracy=70.330882	 loss=1.199898	 lr=0.000100
Epoch[084] Batch [3449]/[3759]	Speed: 70.313112 samples/sec	 accuracy=70.323822	 loss=1.200589	 lr=0.000100
Epoch[084] Batch [3499]/[3759]	Speed: 69.296673 samples/sec	 accuracy=70.315179	 loss=1.200486	 lr=0.000100
Epoch[084] Batch [3549]/[3759]	Speed: 69.429793 samples/sec	 accuracy=70.319982	 loss=1.200274	 lr=0.000100
Epoch[084] Batch [3599]/[3759]	Speed: 68.703014 samples/sec	 accuracy=70.314670	 loss=1.200231	 lr=0.000100
Epoch[084] Batch [3649]/[3759]	Speed: 70.833671 samples/sec	 accuracy=70.297089	 loss=1.201046	 lr=0.000100
Epoch[084] Batch [3699]/[3759]	Speed: 68.350126 samples/sec	 accuracy=70.291807	 loss=1.201308	 lr=0.000100
Epoch[084] Batch [3749]/[3759]	Speed: 78.749847 samples/sec	 accuracy=70.310417	 loss=1.201403	 lr=0.000100
Batch [0049]/[0303]: acc-top1=62.687500 acc-top5=84.250000
Batch [0099]/[0303]: acc-top1=62.828125 acc-top5=84.312500
Batch [0149]/[0303]: acc-top1=63.833333 acc-top5=84.781250
Batch [0199]/[0303]: acc-top1=64.242188 acc-top5=85.093750
Batch [0249]/[0303]: acc-top1=64.543750 acc-top5=85.325000
Batch [0299]/[0303]: acc-top1=64.395833 acc-top5=85.156250
[Epoch 084] training: accuracy=70.310422	 loss=1.201387
[Epoch 084] speed: 69 samples/sec	time cost: 3750.068537
[Epoch 084] validation: acc-top1=64.356436 acc-top5=85.163985 loss=1.555969
Epoch[085] Batch [0049]/[3760]	Speed: 48.598740 samples/sec	 accuracy=70.093750	 loss=1.210484	 lr=0.000100
Epoch[085] Batch [0099]/[3760]	Speed: 67.624233 samples/sec	 accuracy=70.265625	 loss=1.197153	 lr=0.000100
Epoch[085] Batch [0149]/[3760]	Speed: 68.682661 samples/sec	 accuracy=70.302083	 loss=1.190666	 lr=0.000100
Epoch[085] Batch [0199]/[3760]	Speed: 68.548322 samples/sec	 accuracy=70.046875	 loss=1.201949	 lr=0.000100
Epoch[085] Batch [0249]/[3760]	Speed: 69.941067 samples/sec	 accuracy=70.212500	 loss=1.199472	 lr=0.000100
Epoch[085] Batch [0299]/[3760]	Speed: 69.748991 samples/sec	 accuracy=70.359375	 loss=1.199142	 lr=0.000100
Epoch[085] Batch [0349]/[3760]	Speed: 69.861773 samples/sec	 accuracy=70.299107	 loss=1.201401	 lr=0.000100
Epoch[085] Batch [0399]/[3760]	Speed: 71.576712 samples/sec	 accuracy=70.144531	 loss=1.204799	 lr=0.000100
Epoch[085] Batch [0449]/[3760]	Speed: 69.224586 samples/sec	 accuracy=70.031250	 loss=1.210420	 lr=0.000100
Epoch[085] Batch [0499]/[3760]	Speed: 68.809743 samples/sec	 accuracy=69.981250	 loss=1.208151	 lr=0.000100
Epoch[085] Batch [0549]/[3760]	Speed: 70.525849 samples/sec	 accuracy=69.960227	 loss=1.208279	 lr=0.000100
Epoch[085] Batch [0599]/[3760]	Speed: 70.106154 samples/sec	 accuracy=69.963542	 loss=1.206128	 lr=0.000100
Epoch[085] Batch [0649]/[3760]	Speed: 69.537700 samples/sec	 accuracy=69.983173	 loss=1.205749	 lr=0.000100
Epoch[085] Batch [0699]/[3760]	Speed: 69.713507 samples/sec	 accuracy=70.022321	 loss=1.204071	 lr=0.000100
Epoch[085] Batch [0749]/[3760]	Speed: 69.187401 samples/sec	 accuracy=70.110417	 loss=1.201234	 lr=0.000100
Epoch[085] Batch [0799]/[3760]	Speed: 68.163471 samples/sec	 accuracy=70.177734	 loss=1.197789	 lr=0.000100
Epoch[085] Batch [0849]/[3760]	Speed: 68.945332 samples/sec	 accuracy=70.181985	 loss=1.199429	 lr=0.000100
Epoch[085] Batch [0899]/[3760]	Speed: 69.976554 samples/sec	 accuracy=70.161458	 loss=1.200553	 lr=0.000100
Epoch[085] Batch [0949]/[3760]	Speed: 67.443216 samples/sec	 accuracy=70.139803	 loss=1.199671	 lr=0.000100
Epoch[085] Batch [0999]/[3760]	Speed: 72.656480 samples/sec	 accuracy=70.137500	 loss=1.199893	 lr=0.000100
Epoch[085] Batch [1049]/[3760]	Speed: 68.748278 samples/sec	 accuracy=70.055060	 loss=1.202965	 lr=0.000100
Epoch[085] Batch [1099]/[3760]	Speed: 71.425115 samples/sec	 accuracy=70.130682	 loss=1.201903	 lr=0.000100
Epoch[085] Batch [1149]/[3760]	Speed: 68.937537 samples/sec	 accuracy=70.180707	 loss=1.200930	 lr=0.000100
Epoch[085] Batch [1199]/[3760]	Speed: 69.339870 samples/sec	 accuracy=70.210938	 loss=1.198718	 lr=0.000100
Epoch[085] Batch [1249]/[3760]	Speed: 70.637858 samples/sec	 accuracy=70.300000	 loss=1.195976	 lr=0.000100
Epoch[085] Batch [1299]/[3760]	Speed: 69.940917 samples/sec	 accuracy=70.295673	 loss=1.196610	 lr=0.000100
Epoch[085] Batch [1349]/[3760]	Speed: 70.222960 samples/sec	 accuracy=70.292824	 loss=1.196547	 lr=0.000100
Epoch[085] Batch [1399]/[3760]	Speed: 69.934022 samples/sec	 accuracy=70.297991	 loss=1.196015	 lr=0.000100
Epoch[085] Batch [1449]/[3760]	Speed: 69.581752 samples/sec	 accuracy=70.309267	 loss=1.196474	 lr=0.000100
Epoch[085] Batch [1499]/[3760]	Speed: 70.305380 samples/sec	 accuracy=70.303125	 loss=1.196646	 lr=0.000100
Epoch[085] Batch [1549]/[3760]	Speed: 69.728144 samples/sec	 accuracy=70.308468	 loss=1.195991	 lr=0.000100
Epoch[085] Batch [1599]/[3760]	Speed: 69.192537 samples/sec	 accuracy=70.277344	 loss=1.196982	 lr=0.000100
Epoch[085] Batch [1649]/[3760]	Speed: 66.425688 samples/sec	 accuracy=70.281250	 loss=1.196827	 lr=0.000100
Epoch[085] Batch [1699]/[3760]	Speed: 71.409964 samples/sec	 accuracy=70.273897	 loss=1.196697	 lr=0.000100
Epoch[085] Batch [1749]/[3760]	Speed: 70.554491 samples/sec	 accuracy=70.245536	 loss=1.198442	 lr=0.000100
Epoch[085] Batch [1799]/[3760]	Speed: 69.818544 samples/sec	 accuracy=70.236979	 loss=1.199471	 lr=0.000100
Epoch[085] Batch [1849]/[3760]	Speed: 69.365388 samples/sec	 accuracy=70.217905	 loss=1.199786	 lr=0.000100
Epoch[085] Batch [1899]/[3760]	Speed: 69.087670 samples/sec	 accuracy=70.245066	 loss=1.199175	 lr=0.000100
Epoch[085] Batch [1949]/[3760]	Speed: 70.516262 samples/sec	 accuracy=70.204327	 loss=1.200922	 lr=0.000100
Epoch[085] Batch [1999]/[3760]	Speed: 67.166960 samples/sec	 accuracy=70.200781	 loss=1.201649	 lr=0.000100
Epoch[085] Batch [2049]/[3760]	Speed: 72.190593 samples/sec	 accuracy=70.193598	 loss=1.201929	 lr=0.000100
Epoch[085] Batch [2099]/[3760]	Speed: 70.695032 samples/sec	 accuracy=70.206845	 loss=1.201471	 lr=0.000100
Epoch[085] Batch [2149]/[3760]	Speed: 69.039245 samples/sec	 accuracy=70.206395	 loss=1.201598	 lr=0.000100
Epoch[085] Batch [2199]/[3760]	Speed: 70.052830 samples/sec	 accuracy=70.231534	 loss=1.200306	 lr=0.000100
Epoch[085] Batch [2249]/[3760]	Speed: 70.055120 samples/sec	 accuracy=70.258333	 loss=1.199847	 lr=0.000100
Epoch[085] Batch [2299]/[3760]	Speed: 70.091940 samples/sec	 accuracy=70.274457	 loss=1.199103	 lr=0.000100
Epoch[085] Batch [2349]/[3760]	Speed: 69.362579 samples/sec	 accuracy=70.288564	 loss=1.198414	 lr=0.000100
Epoch[085] Batch [2399]/[3760]	Speed: 69.722273 samples/sec	 accuracy=70.294271	 loss=1.198924	 lr=0.000100
Epoch[085] Batch [2449]/[3760]	Speed: 69.538526 samples/sec	 accuracy=70.323342	 loss=1.197709	 lr=0.000100
Epoch[085] Batch [2499]/[3760]	Speed: 68.753845 samples/sec	 accuracy=70.338750	 loss=1.197519	 lr=0.000100
Epoch[085] Batch [2549]/[3760]	Speed: 70.395480 samples/sec	 accuracy=70.324755	 loss=1.197690	 lr=0.000100
Epoch[085] Batch [2599]/[3760]	Speed: 67.739205 samples/sec	 accuracy=70.340745	 loss=1.197743	 lr=0.000100
Epoch[085] Batch [2649]/[3760]	Speed: 70.523575 samples/sec	 accuracy=70.342571	 loss=1.198217	 lr=0.000100
Epoch[085] Batch [2699]/[3760]	Speed: 70.042880 samples/sec	 accuracy=70.328704	 loss=1.198314	 lr=0.000100
Epoch[085] Batch [2749]/[3760]	Speed: 69.610686 samples/sec	 accuracy=70.344886	 loss=1.198169	 lr=0.000100
Epoch[085] Batch [2799]/[3760]	Speed: 69.654987 samples/sec	 accuracy=70.357701	 loss=1.197681	 lr=0.000100
Epoch[085] Batch [2849]/[3760]	Speed: 69.412709 samples/sec	 accuracy=70.351425	 loss=1.198252	 lr=0.000100
Epoch[085] Batch [2899]/[3760]	Speed: 68.563063 samples/sec	 accuracy=70.365302	 loss=1.198299	 lr=0.000100
Epoch[085] Batch [2949]/[3760]	Speed: 70.871661 samples/sec	 accuracy=70.349047	 loss=1.198292	 lr=0.000100
Epoch[085] Batch [2999]/[3760]	Speed: 67.825288 samples/sec	 accuracy=70.357813	 loss=1.198107	 lr=0.000100
Epoch[085] Batch [3049]/[3760]	Speed: 70.892756 samples/sec	 accuracy=70.360143	 loss=1.198298	 lr=0.000100
Epoch[085] Batch [3099]/[3760]	Speed: 69.525447 samples/sec	 accuracy=70.366431	 loss=1.197849	 lr=0.000100
Epoch[085] Batch [3149]/[3760]	Speed: 69.520760 samples/sec	 accuracy=70.380456	 loss=1.197527	 lr=0.000100
Epoch[085] Batch [3199]/[3760]	Speed: 69.550829 samples/sec	 accuracy=70.364258	 loss=1.198017	 lr=0.000100
Epoch[085] Batch [3249]/[3760]	Speed: 69.494338 samples/sec	 accuracy=70.365385	 loss=1.197773	 lr=0.000100
Epoch[085] Batch [3299]/[3760]	Speed: 70.124904 samples/sec	 accuracy=70.355587	 loss=1.197831	 lr=0.000100
Epoch[085] Batch [3349]/[3760]	Speed: 70.231959 samples/sec	 accuracy=70.347948	 loss=1.197901	 lr=0.000100
Epoch[085] Batch [3399]/[3760]	Speed: 69.399283 samples/sec	 accuracy=70.353401	 loss=1.197994	 lr=0.000100
Epoch[085] Batch [3449]/[3760]	Speed: 68.377793 samples/sec	 accuracy=70.355978	 loss=1.197811	 lr=0.000100
Epoch[085] Batch [3499]/[3760]	Speed: 69.255147 samples/sec	 accuracy=70.347768	 loss=1.198711	 lr=0.000100
Epoch[085] Batch [3549]/[3760]	Speed: 70.037399 samples/sec	 accuracy=70.348151	 loss=1.198940	 lr=0.000100
Epoch[085] Batch [3599]/[3760]	Speed: 67.452276 samples/sec	 accuracy=70.353299	 loss=1.198710	 lr=0.000100
Epoch[085] Batch [3649]/[3760]	Speed: 71.307788 samples/sec	 accuracy=70.354880	 loss=1.198622	 lr=0.000100
Epoch[085] Batch [3699]/[3760]	Speed: 70.756974 samples/sec	 accuracy=70.346706	 loss=1.199185	 lr=0.000100
Epoch[085] Batch [3749]/[3760]	Speed: 76.745149 samples/sec	 accuracy=70.340417	 loss=1.199619	 lr=0.000100
Batch [0049]/[0303]: acc-top1=62.687500 acc-top5=84.031250
Batch [0099]/[0303]: acc-top1=62.812500 acc-top5=84.218750
Batch [0149]/[0303]: acc-top1=63.833333 acc-top5=84.541667
Batch [0199]/[0303]: acc-top1=64.273438 acc-top5=84.890625
Batch [0249]/[0303]: acc-top1=64.625000 acc-top5=85.187500
Batch [0299]/[0303]: acc-top1=64.442708 acc-top5=85.015625
[Epoch 085] training: accuracy=70.345329	 loss=1.199452
[Epoch 085] speed: 69 samples/sec	time cost: 3755.326121
[Epoch 085] validation: acc-top1=64.413160 acc-top5=85.029909 loss=1.560570
Epoch[086] Batch [0049]/[3760]	Speed: 48.038606 samples/sec	 accuracy=70.093750	 loss=1.196875	 lr=0.000100
Epoch[086] Batch [0099]/[3760]	Speed: 69.808278 samples/sec	 accuracy=70.343750	 loss=1.186314	 lr=0.000100
Epoch[086] Batch [0149]/[3760]	Speed: 67.905801 samples/sec	 accuracy=70.395833	 loss=1.189997	 lr=0.000100
Epoch[086] Batch [0199]/[3760]	Speed: 71.433896 samples/sec	 accuracy=70.539062	 loss=1.181310	 lr=0.000100
Epoch[086] Batch [0249]/[3760]	Speed: 69.679499 samples/sec	 accuracy=70.443750	 loss=1.189860	 lr=0.000100
Epoch[086] Batch [0299]/[3760]	Speed: 70.695808 samples/sec	 accuracy=70.645833	 loss=1.190271	 lr=0.000100
Epoch[086] Batch [0349]/[3760]	Speed: 68.806163 samples/sec	 accuracy=70.750000	 loss=1.191245	 lr=0.000100
Epoch[086] Batch [0399]/[3760]	Speed: 69.319392 samples/sec	 accuracy=70.675781	 loss=1.192257	 lr=0.000100
Epoch[086] Batch [0449]/[3760]	Speed: 68.653833 samples/sec	 accuracy=70.635417	 loss=1.191787	 lr=0.000100
Epoch[086] Batch [0499]/[3760]	Speed: 70.528316 samples/sec	 accuracy=70.625000	 loss=1.194615	 lr=0.000100
Epoch[086] Batch [0549]/[3760]	Speed: 66.156794 samples/sec	 accuracy=70.625000	 loss=1.194087	 lr=0.000100
Epoch[086] Batch [0599]/[3760]	Speed: 72.430904 samples/sec	 accuracy=70.682292	 loss=1.193869	 lr=0.000100
Epoch[086] Batch [0649]/[3760]	Speed: 70.300056 samples/sec	 accuracy=70.603365	 loss=1.195781	 lr=0.000100
Epoch[086] Batch [0699]/[3760]	Speed: 68.999045 samples/sec	 accuracy=70.645089	 loss=1.196120	 lr=0.000100
Epoch[086] Batch [0749]/[3760]	Speed: 70.028260 samples/sec	 accuracy=70.616667	 loss=1.195635	 lr=0.000100
Epoch[086] Batch [0799]/[3760]	Speed: 67.945458 samples/sec	 accuracy=70.642578	 loss=1.192931	 lr=0.000100
Epoch[086] Batch [0849]/[3760]	Speed: 70.768690 samples/sec	 accuracy=70.615809	 loss=1.193903	 lr=0.000100
Epoch[086] Batch [0899]/[3760]	Speed: 69.590928 samples/sec	 accuracy=70.583333	 loss=1.195548	 lr=0.000100
Epoch[086] Batch [0949]/[3760]	Speed: 69.339312 samples/sec	 accuracy=70.490132	 loss=1.198092	 lr=0.000100
Epoch[086] Batch [0999]/[3760]	Speed: 69.617882 samples/sec	 accuracy=70.485937	 loss=1.197874	 lr=0.000100
Epoch[086] Batch [1049]/[3760]	Speed: 68.741589 samples/sec	 accuracy=70.480655	 loss=1.197918	 lr=0.000100
Epoch[086] Batch [1099]/[3760]	Speed: 70.386959 samples/sec	 accuracy=70.450284	 loss=1.200114	 lr=0.000100
Epoch[086] Batch [1149]/[3760]	Speed: 68.033124 samples/sec	 accuracy=70.410326	 loss=1.201257	 lr=0.000100
Epoch[086] Batch [1199]/[3760]	Speed: 70.748806 samples/sec	 accuracy=70.415365	 loss=1.200895	 lr=0.000100
Epoch[086] Batch [1249]/[3760]	Speed: 69.082237 samples/sec	 accuracy=70.402500	 loss=1.202066	 lr=0.000100
Epoch[086] Batch [1299]/[3760]	Speed: 70.087485 samples/sec	 accuracy=70.396635	 loss=1.202247	 lr=0.000100
Epoch[086] Batch [1349]/[3760]	Speed: 68.664470 samples/sec	 accuracy=70.361111	 loss=1.203519	 lr=0.000100
Epoch[086] Batch [1399]/[3760]	Speed: 69.813448 samples/sec	 accuracy=70.399554	 loss=1.202755	 lr=0.000100
Epoch[086] Batch [1449]/[3760]	Speed: 70.730356 samples/sec	 accuracy=70.400862	 loss=1.203081	 lr=0.000100
Epoch[086] Batch [1499]/[3760]	Speed: 68.864882 samples/sec	 accuracy=70.438542	 loss=1.202124	 lr=0.000100
Epoch[086] Batch [1549]/[3760]	Speed: 68.290661 samples/sec	 accuracy=70.431452	 loss=1.201925	 lr=0.000100
Epoch[086] Batch [1599]/[3760]	Speed: 70.234231 samples/sec	 accuracy=70.457031	 loss=1.199502	 lr=0.000100
Epoch[086] Batch [1649]/[3760]	Speed: 70.002294 samples/sec	 accuracy=70.429924	 loss=1.200418	 lr=0.000100
Epoch[086] Batch [1699]/[3760]	Speed: 68.865385 samples/sec	 accuracy=70.420037	 loss=1.200057	 lr=0.000100
Epoch[086] Batch [1749]/[3760]	Speed: 66.216969 samples/sec	 accuracy=70.430357	 loss=1.200130	 lr=0.000100
Epoch[086] Batch [1799]/[3760]	Speed: 73.510729 samples/sec	 accuracy=70.460069	 loss=1.197914	 lr=0.000100
Epoch[086] Batch [1849]/[3760]	Speed: 69.958824 samples/sec	 accuracy=70.467061	 loss=1.197638	 lr=0.000100
Epoch[086] Batch [1899]/[3760]	Speed: 70.105195 samples/sec	 accuracy=70.466283	 loss=1.197762	 lr=0.000100
Epoch[086] Batch [1949]/[3760]	Speed: 69.645255 samples/sec	 accuracy=70.473558	 loss=1.196676	 lr=0.000100
Epoch[086] Batch [1999]/[3760]	Speed: 70.046617 samples/sec	 accuracy=70.472656	 loss=1.197544	 lr=0.000100
Epoch[086] Batch [2049]/[3760]	Speed: 68.246778 samples/sec	 accuracy=70.468750	 loss=1.198577	 lr=0.000100
Epoch[086] Batch [2099]/[3760]	Speed: 70.274055 samples/sec	 accuracy=70.459821	 loss=1.199358	 lr=0.000100
Epoch[086] Batch [2149]/[3760]	Speed: 70.206686 samples/sec	 accuracy=70.470203	 loss=1.198946	 lr=0.000100
Epoch[086] Batch [2199]/[3760]	Speed: 70.308873 samples/sec	 accuracy=70.459517	 loss=1.199345	 lr=0.000100
Epoch[086] Batch [2249]/[3760]	Speed: 68.818224 samples/sec	 accuracy=70.459722	 loss=1.199390	 lr=0.000100
Epoch[086] Batch [2299]/[3760]	Speed: 70.625214 samples/sec	 accuracy=70.467391	 loss=1.198657	 lr=0.000100
Epoch[086] Batch [2349]/[3760]	Speed: 70.183359 samples/sec	 accuracy=70.471410	 loss=1.198896	 lr=0.000100
Epoch[086] Batch [2399]/[3760]	Speed: 68.404223 samples/sec	 accuracy=70.445964	 loss=1.199579	 lr=0.000100
Epoch[086] Batch [2449]/[3760]	Speed: 70.647850 samples/sec	 accuracy=70.419643	 loss=1.200214	 lr=0.000100
Epoch[086] Batch [2499]/[3760]	Speed: 70.635645 samples/sec	 accuracy=70.408750	 loss=1.200730	 lr=0.000100
Epoch[086] Batch [2549]/[3760]	Speed: 69.614780 samples/sec	 accuracy=70.420343	 loss=1.200534	 lr=0.000100
Epoch[086] Batch [2599]/[3760]	Speed: 68.786182 samples/sec	 accuracy=70.425481	 loss=1.200375	 lr=0.000100
Epoch[086] Batch [2649]/[3760]	Speed: 70.257491 samples/sec	 accuracy=70.432783	 loss=1.199613	 lr=0.000100
Epoch[086] Batch [2699]/[3760]	Speed: 69.827101 samples/sec	 accuracy=70.439236	 loss=1.199565	 lr=0.000100
Epoch[086] Batch [2749]/[3760]	Speed: 68.551715 samples/sec	 accuracy=70.445455	 loss=1.199765	 lr=0.000100
Epoch[086] Batch [2799]/[3760]	Speed: 70.552841 samples/sec	 accuracy=70.444196	 loss=1.199352	 lr=0.000100
Epoch[086] Batch [2849]/[3760]	Speed: 69.489336 samples/sec	 accuracy=70.450110	 loss=1.199137	 lr=0.000100
Epoch[086] Batch [2899]/[3760]	Speed: 69.199673 samples/sec	 accuracy=70.460129	 loss=1.198770	 lr=0.000100
Epoch[086] Batch [2949]/[3760]	Speed: 70.724019 samples/sec	 accuracy=70.461335	 loss=1.198686	 lr=0.000100
Epoch[086] Batch [2999]/[3760]	Speed: 69.788705 samples/sec	 accuracy=70.460938	 loss=1.198771	 lr=0.000100
Epoch[086] Batch [3049]/[3760]	Speed: 70.374270 samples/sec	 accuracy=70.451332	 loss=1.199196	 lr=0.000100
Epoch[086] Batch [3099]/[3760]	Speed: 69.638695 samples/sec	 accuracy=70.462198	 loss=1.198766	 lr=0.000100
Epoch[086] Batch [3149]/[3760]	Speed: 68.544942 samples/sec	 accuracy=70.439980	 loss=1.199829	 lr=0.000100
Epoch[086] Batch [3199]/[3760]	Speed: 70.649325 samples/sec	 accuracy=70.464844	 loss=1.199221	 lr=0.000100
Epoch[086] Batch [3249]/[3760]	Speed: 68.088333 samples/sec	 accuracy=70.489423	 loss=1.198183	 lr=0.000100
Epoch[086] Batch [3299]/[3760]	Speed: 70.559116 samples/sec	 accuracy=70.487689	 loss=1.198041	 lr=0.000100
Epoch[086] Batch [3349]/[3760]	Speed: 69.419782 samples/sec	 accuracy=70.481810	 loss=1.198217	 lr=0.000100
Epoch[086] Batch [3399]/[3760]	Speed: 70.306955 samples/sec	 accuracy=70.469210	 loss=1.198305	 lr=0.000100
Epoch[086] Batch [3449]/[3760]	Speed: 69.882014 samples/sec	 accuracy=70.457880	 loss=1.198760	 lr=0.000100
Epoch[086] Batch [3499]/[3760]	Speed: 70.272961 samples/sec	 accuracy=70.441071	 loss=1.199774	 lr=0.000100
Epoch[086] Batch [3549]/[3760]	Speed: 69.801394 samples/sec	 accuracy=70.440141	 loss=1.199863	 lr=0.000100
Epoch[086] Batch [3599]/[3760]	Speed: 70.421395 samples/sec	 accuracy=70.441406	 loss=1.200090	 lr=0.000100
Epoch[086] Batch [3649]/[3760]	Speed: 67.275372 samples/sec	 accuracy=70.460188	 loss=1.199341	 lr=0.000100
Epoch[086] Batch [3699]/[3760]	Speed: 70.846054 samples/sec	 accuracy=70.458193	 loss=1.199062	 lr=0.000100
Epoch[086] Batch [3749]/[3760]	Speed: 77.557239 samples/sec	 accuracy=70.448750	 loss=1.199811	 lr=0.000100
Batch [0049]/[0303]: acc-top1=62.843750 acc-top5=83.906250
Batch [0099]/[0303]: acc-top1=62.593750 acc-top5=84.250000
Batch [0149]/[0303]: acc-top1=63.625000 acc-top5=84.677083
Batch [0199]/[0303]: acc-top1=64.203125 acc-top5=85.000000
Batch [0249]/[0303]: acc-top1=64.437500 acc-top5=85.225000
Batch [0299]/[0303]: acc-top1=64.333333 acc-top5=85.046875
[Epoch 086] training: accuracy=70.457530	 loss=1.199636
[Epoch 086] speed: 69 samples/sec	time cost: 3752.739240
[Epoch 086] validation: acc-top1=64.299711 acc-top5=85.045380 loss=1.555606
Epoch[087] Batch [0049]/[3759]	Speed: 48.343040 samples/sec	 accuracy=70.281250	 loss=1.183364	 lr=0.000100
Epoch[087] Batch [0099]/[3759]	Speed: 67.543825 samples/sec	 accuracy=70.484375	 loss=1.190897	 lr=0.000100
Epoch[087] Batch [0149]/[3759]	Speed: 69.959490 samples/sec	 accuracy=70.354167	 loss=1.185899	 lr=0.000100
Epoch[087] Batch [0199]/[3759]	Speed: 68.358653 samples/sec	 accuracy=70.609375	 loss=1.179009	 lr=0.000100
Epoch[087] Batch [0249]/[3759]	Speed: 71.877948 samples/sec	 accuracy=70.600000	 loss=1.182956	 lr=0.000100
Epoch[087] Batch [0299]/[3759]	Speed: 67.523001 samples/sec	 accuracy=70.552083	 loss=1.179012	 lr=0.000100
Epoch[087] Batch [0349]/[3759]	Speed: 71.370957 samples/sec	 accuracy=70.424107	 loss=1.182098	 lr=0.000100
Epoch[087] Batch [0399]/[3759]	Speed: 70.699835 samples/sec	 accuracy=70.347656	 loss=1.188545	 lr=0.000100
Epoch[087] Batch [0449]/[3759]	Speed: 66.689560 samples/sec	 accuracy=70.215278	 loss=1.195264	 lr=0.000100
Epoch[087] Batch [0499]/[3759]	Speed: 72.201794 samples/sec	 accuracy=70.284375	 loss=1.194975	 lr=0.000100
Epoch[087] Batch [0549]/[3759]	Speed: 68.937009 samples/sec	 accuracy=70.428977	 loss=1.190212	 lr=0.000100
Epoch[087] Batch [0599]/[3759]	Speed: 70.550346 samples/sec	 accuracy=70.565104	 loss=1.183489	 lr=0.000100
Epoch[087] Batch [0649]/[3759]	Speed: 70.270167 samples/sec	 accuracy=70.509615	 loss=1.186536	 lr=0.000100
Epoch[087] Batch [0699]/[3759]	Speed: 69.109622 samples/sec	 accuracy=70.546875	 loss=1.185695	 lr=0.000100
Epoch[087] Batch [0749]/[3759]	Speed: 70.339867 samples/sec	 accuracy=70.662500	 loss=1.184492	 lr=0.000100
Epoch[087] Batch [0799]/[3759]	Speed: 69.139507 samples/sec	 accuracy=70.699219	 loss=1.182406	 lr=0.000100
Epoch[087] Batch [0849]/[3759]	Speed: 69.568683 samples/sec	 accuracy=70.628676	 loss=1.185716	 lr=0.000100
Epoch[087] Batch [0899]/[3759]	Speed: 70.200236 samples/sec	 accuracy=70.569444	 loss=1.187083	 lr=0.000100
Epoch[087] Batch [0949]/[3759]	Speed: 69.660117 samples/sec	 accuracy=70.595395	 loss=1.188030	 lr=0.000100
Epoch[087] Batch [0999]/[3759]	Speed: 70.270912 samples/sec	 accuracy=70.604687	 loss=1.186719	 lr=0.000100
Epoch[087] Batch [1049]/[3759]	Speed: 70.207440 samples/sec	 accuracy=70.563988	 loss=1.188175	 lr=0.000100
Epoch[087] Batch [1099]/[3759]	Speed: 69.338726 samples/sec	 accuracy=70.478693	 loss=1.191000	 lr=0.000100
Epoch[087] Batch [1149]/[3759]	Speed: 70.174770 samples/sec	 accuracy=70.437500	 loss=1.192478	 lr=0.000100
Epoch[087] Batch [1199]/[3759]	Speed: 70.095593 samples/sec	 accuracy=70.399740	 loss=1.194612	 lr=0.000100
Epoch[087] Batch [1249]/[3759]	Speed: 69.722857 samples/sec	 accuracy=70.405000	 loss=1.195443	 lr=0.000100
Epoch[087] Batch [1299]/[3759]	Speed: 70.296071 samples/sec	 accuracy=70.418269	 loss=1.195495	 lr=0.000100
Epoch[087] Batch [1349]/[3759]	Speed: 70.064161 samples/sec	 accuracy=70.428241	 loss=1.195126	 lr=0.000100
Epoch[087] Batch [1399]/[3759]	Speed: 68.909623 samples/sec	 accuracy=70.466518	 loss=1.194314	 lr=0.000100
Epoch[087] Batch [1449]/[3759]	Speed: 69.938667 samples/sec	 accuracy=70.448276	 loss=1.194870	 lr=0.000100
Epoch[087] Batch [1499]/[3759]	Speed: 70.676204 samples/sec	 accuracy=70.501042	 loss=1.193329	 lr=0.000100
Epoch[087] Batch [1549]/[3759]	Speed: 69.048032 samples/sec	 accuracy=70.533266	 loss=1.192572	 lr=0.000100
Epoch[087] Batch [1599]/[3759]	Speed: 69.303788 samples/sec	 accuracy=70.502930	 loss=1.193545	 lr=0.000100
Epoch[087] Batch [1649]/[3759]	Speed: 70.319743 samples/sec	 accuracy=70.525568	 loss=1.192125	 lr=0.000100
Epoch[087] Batch [1699]/[3759]	Speed: 68.921981 samples/sec	 accuracy=70.556985	 loss=1.191666	 lr=0.000100
Epoch[087] Batch [1749]/[3759]	Speed: 70.532556 samples/sec	 accuracy=70.571429	 loss=1.191372	 lr=0.000100
Epoch[087] Batch [1799]/[3759]	Speed: 70.049647 samples/sec	 accuracy=70.554688	 loss=1.192026	 lr=0.000100
Epoch[087] Batch [1849]/[3759]	Speed: 69.982746 samples/sec	 accuracy=70.576858	 loss=1.192455	 lr=0.000100
Epoch[087] Batch [1899]/[3759]	Speed: 69.757094 samples/sec	 accuracy=70.564967	 loss=1.193049	 lr=0.000100
Epoch[087] Batch [1949]/[3759]	Speed: 69.421710 samples/sec	 accuracy=70.566506	 loss=1.192833	 lr=0.000100
Epoch[087] Batch [1999]/[3759]	Speed: 71.529845 samples/sec	 accuracy=70.589062	 loss=1.191727	 lr=0.000100
Epoch[087] Batch [2049]/[3759]	Speed: 67.420679 samples/sec	 accuracy=70.617378	 loss=1.191264	 lr=0.000100
Epoch[087] Batch [2099]/[3759]	Speed: 70.570173 samples/sec	 accuracy=70.604911	 loss=1.191931	 lr=0.000100
Epoch[087] Batch [2149]/[3759]	Speed: 69.627341 samples/sec	 accuracy=70.622820	 loss=1.191213	 lr=0.000100
Epoch[087] Batch [2199]/[3759]	Speed: 70.748768 samples/sec	 accuracy=70.637784	 loss=1.191094	 lr=0.000100
Epoch[087] Batch [2249]/[3759]	Speed: 68.560112 samples/sec	 accuracy=70.629167	 loss=1.191564	 lr=0.000100
Epoch[087] Batch [2299]/[3759]	Speed: 70.622633 samples/sec	 accuracy=70.657609	 loss=1.190661	 lr=0.000100
Epoch[087] Batch [2349]/[3759]	Speed: 69.487240 samples/sec	 accuracy=70.642952	 loss=1.190866	 lr=0.000100
Epoch[087] Batch [2399]/[3759]	Speed: 69.795091 samples/sec	 accuracy=70.667318	 loss=1.189869	 lr=0.000100
Epoch[087] Batch [2449]/[3759]	Speed: 70.313162 samples/sec	 accuracy=70.670918	 loss=1.189276	 lr=0.000100
Epoch[087] Batch [2499]/[3759]	Speed: 69.978924 samples/sec	 accuracy=70.638125	 loss=1.190258	 lr=0.000100
Epoch[087] Batch [2549]/[3759]	Speed: 69.995168 samples/sec	 accuracy=70.635417	 loss=1.190449	 lr=0.000100
Epoch[087] Batch [2599]/[3759]	Speed: 69.080796 samples/sec	 accuracy=70.638221	 loss=1.190016	 lr=0.000100
Epoch[087] Batch [2649]/[3759]	Speed: 66.711098 samples/sec	 accuracy=70.686910	 loss=1.188255	 lr=0.000100
Epoch[087] Batch [2699]/[3759]	Speed: 71.962364 samples/sec	 accuracy=70.689236	 loss=1.188366	 lr=0.000100
Epoch[087] Batch [2749]/[3759]	Speed: 71.005736 samples/sec	 accuracy=70.700000	 loss=1.187965	 lr=0.000100
Epoch[087] Batch [2799]/[3759]	Speed: 69.794049 samples/sec	 accuracy=70.676339	 loss=1.188918	 lr=0.000100
Epoch[087] Batch [2849]/[3759]	Speed: 69.206162 samples/sec	 accuracy=70.664474	 loss=1.189195	 lr=0.000100
Epoch[087] Batch [2899]/[3759]	Speed: 69.358959 samples/sec	 accuracy=70.661638	 loss=1.189067	 lr=0.000100
Epoch[087] Batch [2949]/[3759]	Speed: 70.224455 samples/sec	 accuracy=70.649894	 loss=1.189558	 lr=0.000100
Epoch[087] Batch [2999]/[3759]	Speed: 68.078214 samples/sec	 accuracy=70.680729	 loss=1.188801	 lr=0.000100
Epoch[087] Batch [3049]/[3759]	Speed: 70.908386 samples/sec	 accuracy=70.665471	 loss=1.189267	 lr=0.000100
Epoch[087] Batch [3099]/[3759]	Speed: 69.476173 samples/sec	 accuracy=70.645665	 loss=1.189407	 lr=0.000100
Epoch[087] Batch [3149]/[3759]	Speed: 70.056365 samples/sec	 accuracy=70.656250	 loss=1.188465	 lr=0.000100
Epoch[087] Batch [3199]/[3759]	Speed: 70.017445 samples/sec	 accuracy=70.629883	 loss=1.189628	 lr=0.000100
Epoch[087] Batch [3249]/[3759]	Speed: 69.041087 samples/sec	 accuracy=70.636058	 loss=1.189844	 lr=0.000100
Epoch[087] Batch [3299]/[3759]	Speed: 69.681361 samples/sec	 accuracy=70.614110	 loss=1.190400	 lr=0.000100
Epoch[087] Batch [3349]/[3759]	Speed: 70.808580 samples/sec	 accuracy=70.595616	 loss=1.190612	 lr=0.000100
Epoch[087] Batch [3399]/[3759]	Speed: 67.548191 samples/sec	 accuracy=70.589154	 loss=1.190674	 lr=0.000100
Epoch[087] Batch [3449]/[3759]	Speed: 71.059507 samples/sec	 accuracy=70.586957	 loss=1.189951	 lr=0.000100
Epoch[087] Batch [3499]/[3759]	Speed: 69.904963 samples/sec	 accuracy=70.570089	 loss=1.190390	 lr=0.000100
Epoch[087] Batch [3549]/[3759]	Speed: 67.510168 samples/sec	 accuracy=70.572183	 loss=1.190281	 lr=0.000100
Epoch[087] Batch [3599]/[3759]	Speed: 71.879480 samples/sec	 accuracy=70.585069	 loss=1.189756	 lr=0.000100
Epoch[087] Batch [3649]/[3759]	Speed: 69.087954 samples/sec	 accuracy=70.579623	 loss=1.189712	 lr=0.000100
Epoch[087] Batch [3699]/[3759]	Speed: 70.749921 samples/sec	 accuracy=70.571368	 loss=1.190119	 lr=0.000100
Epoch[087] Batch [3749]/[3759]	Speed: 78.434942 samples/sec	 accuracy=70.577917	 loss=1.189920	 lr=0.000100
Batch [0049]/[0303]: acc-top1=62.812500 acc-top5=84.375000
Batch [0099]/[0303]: acc-top1=62.921875 acc-top5=84.500000
Batch [0149]/[0303]: acc-top1=63.875000 acc-top5=84.906250
Batch [0199]/[0303]: acc-top1=64.320312 acc-top5=85.195312
Batch [0249]/[0303]: acc-top1=64.525000 acc-top5=85.368750
Batch [0299]/[0303]: acc-top1=64.328125 acc-top5=85.197917
[Epoch 087] training: accuracy=70.581438	 loss=1.189735
[Epoch 087] speed: 69 samples/sec	time cost: 3747.545294
[Epoch 087] validation: acc-top1=64.304868 acc-top5=85.200083 loss=1.551887
Epoch[088] Batch [0049]/[3760]	Speed: 48.199429 samples/sec	 accuracy=69.062500	 loss=1.251704	 lr=0.000100
Epoch[088] Batch [0099]/[3760]	Speed: 69.401103 samples/sec	 accuracy=69.703125	 loss=1.216190	 lr=0.000100
Epoch[088] Batch [0149]/[3760]	Speed: 67.536231 samples/sec	 accuracy=69.906250	 loss=1.208041	 lr=0.000100
Epoch[088] Batch [0199]/[3760]	Speed: 68.757982 samples/sec	 accuracy=69.898438	 loss=1.205029	 lr=0.000100
Epoch[088] Batch [0249]/[3760]	Speed: 70.265135 samples/sec	 accuracy=69.918750	 loss=1.213321	 lr=0.000100
Epoch[088] Batch [0299]/[3760]	Speed: 71.220560 samples/sec	 accuracy=69.937500	 loss=1.212224	 lr=0.000100
Epoch[088] Batch [0349]/[3760]	Speed: 67.951759 samples/sec	 accuracy=70.102679	 loss=1.206062	 lr=0.000100
Epoch[088] Batch [0399]/[3760]	Speed: 70.307226 samples/sec	 accuracy=70.019531	 loss=1.213661	 lr=0.000100
Epoch[088] Batch [0449]/[3760]	Speed: 70.430921 samples/sec	 accuracy=70.003472	 loss=1.214364	 lr=0.000100
Epoch[088] Batch [0499]/[3760]	Speed: 70.213833 samples/sec	 accuracy=70.121875	 loss=1.206265	 lr=0.000100
Epoch[088] Batch [0549]/[3760]	Speed: 69.643689 samples/sec	 accuracy=69.991477	 loss=1.211752	 lr=0.000100
Epoch[088] Batch [0599]/[3760]	Speed: 69.514363 samples/sec	 accuracy=70.000000	 loss=1.213297	 lr=0.000100
Epoch[088] Batch [0649]/[3760]	Speed: 70.313470 samples/sec	 accuracy=70.052885	 loss=1.211745	 lr=0.000100
Epoch[088] Batch [0699]/[3760]	Speed: 70.074346 samples/sec	 accuracy=70.145089	 loss=1.209536	 lr=0.000100
Epoch[088] Batch [0749]/[3760]	Speed: 69.005007 samples/sec	 accuracy=70.216667	 loss=1.208363	 lr=0.000100
Epoch[088] Batch [0799]/[3760]	Speed: 68.921454 samples/sec	 accuracy=70.162109	 loss=1.210176	 lr=0.000100
Epoch[088] Batch [0849]/[3760]	Speed: 68.053451 samples/sec	 accuracy=70.161765	 loss=1.210165	 lr=0.000100
Epoch[088] Batch [0899]/[3760]	Speed: 72.376632 samples/sec	 accuracy=70.144097	 loss=1.209457	 lr=0.000100
Epoch[088] Batch [0949]/[3760]	Speed: 70.267644 samples/sec	 accuracy=70.162829	 loss=1.208629	 lr=0.000100
Epoch[088] Batch [0999]/[3760]	Speed: 69.741824 samples/sec	 accuracy=70.182812	 loss=1.208327	 lr=0.000100
Epoch[088] Batch [1049]/[3760]	Speed: 67.823885 samples/sec	 accuracy=70.175595	 loss=1.207622	 lr=0.000100
Epoch[088] Batch [1099]/[3760]	Speed: 70.616045 samples/sec	 accuracy=70.176136	 loss=1.206662	 lr=0.000100
Epoch[088] Batch [1149]/[3760]	Speed: 67.708891 samples/sec	 accuracy=70.202446	 loss=1.205824	 lr=0.000100
Epoch[088] Batch [1199]/[3760]	Speed: 72.445185 samples/sec	 accuracy=70.190104	 loss=1.205637	 lr=0.000100
Epoch[088] Batch [1249]/[3760]	Speed: 69.502632 samples/sec	 accuracy=70.226250	 loss=1.204159	 lr=0.000100
Epoch[088] Batch [1299]/[3760]	Speed: 70.814040 samples/sec	 accuracy=70.229567	 loss=1.203250	 lr=0.000100
Epoch[088] Batch [1349]/[3760]	Speed: 69.058110 samples/sec	 accuracy=70.199074	 loss=1.202475	 lr=0.000100
Epoch[088] Batch [1399]/[3760]	Speed: 69.930616 samples/sec	 accuracy=70.195312	 loss=1.202789	 lr=0.000100
Epoch[088] Batch [1449]/[3760]	Speed: 69.498716 samples/sec	 accuracy=70.202586	 loss=1.201987	 lr=0.000100
Epoch[088] Batch [1499]/[3760]	Speed: 66.017232 samples/sec	 accuracy=70.186458	 loss=1.202182	 lr=0.000100
Epoch[088] Batch [1549]/[3760]	Speed: 71.501192 samples/sec	 accuracy=70.200605	 loss=1.202173	 lr=0.000100
Epoch[088] Batch [1599]/[3760]	Speed: 71.157162 samples/sec	 accuracy=70.235352	 loss=1.200815	 lr=0.000100
Epoch[088] Batch [1649]/[3760]	Speed: 70.369111 samples/sec	 accuracy=70.245265	 loss=1.201452	 lr=0.000100
Epoch[088] Batch [1699]/[3760]	Speed: 67.839803 samples/sec	 accuracy=70.242647	 loss=1.201130	 lr=0.000100
Epoch[088] Batch [1749]/[3760]	Speed: 69.722472 samples/sec	 accuracy=70.260714	 loss=1.200673	 lr=0.000100
Epoch[088] Batch [1799]/[3760]	Speed: 70.470191 samples/sec	 accuracy=70.271701	 loss=1.200420	 lr=0.000100
Epoch[088] Batch [1849]/[3760]	Speed: 69.269015 samples/sec	 accuracy=70.266047	 loss=1.200560	 lr=0.000100
Epoch[088] Batch [1899]/[3760]	Speed: 66.795709 samples/sec	 accuracy=70.263158	 loss=1.200778	 lr=0.000100
Epoch[088] Batch [1949]/[3760]	Speed: 71.492083 samples/sec	 accuracy=70.275641	 loss=1.200352	 lr=0.000100
Epoch[088] Batch [1999]/[3760]	Speed: 66.639728 samples/sec	 accuracy=70.289062	 loss=1.200665	 lr=0.000100
Epoch[088] Batch [2049]/[3760]	Speed: 71.548616 samples/sec	 accuracy=70.301067	 loss=1.200419	 lr=0.000100
Epoch[088] Batch [2099]/[3760]	Speed: 70.624130 samples/sec	 accuracy=70.313244	 loss=1.200859	 lr=0.000100
Epoch[088] Batch [2149]/[3760]	Speed: 68.788638 samples/sec	 accuracy=70.313227	 loss=1.201503	 lr=0.000100
Epoch[088] Batch [2199]/[3760]	Speed: 71.303455 samples/sec	 accuracy=70.321023	 loss=1.201233	 lr=0.000100
Epoch[088] Batch [2249]/[3760]	Speed: 69.141379 samples/sec	 accuracy=70.323611	 loss=1.201346	 lr=0.000100
Epoch[088] Batch [2299]/[3760]	Speed: 70.022268 samples/sec	 accuracy=70.321332	 loss=1.201838	 lr=0.000100
Epoch[088] Batch [2349]/[3760]	Speed: 69.197664 samples/sec	 accuracy=70.305851	 loss=1.201836	 lr=0.000100
Epoch[088] Batch [2399]/[3760]	Speed: 68.745924 samples/sec	 accuracy=70.315104	 loss=1.201673	 lr=0.000100
Epoch[088] Batch [2449]/[3760]	Speed: 70.038245 samples/sec	 accuracy=70.348214	 loss=1.200442	 lr=0.000100
Epoch[088] Batch [2499]/[3760]	Speed: 69.968138 samples/sec	 accuracy=70.343750	 loss=1.200625	 lr=0.000100
Epoch[088] Batch [2549]/[3760]	Speed: 69.902571 samples/sec	 accuracy=70.375000	 loss=1.199838	 lr=0.000100
Epoch[088] Batch [2599]/[3760]	Speed: 69.208682 samples/sec	 accuracy=70.388822	 loss=1.199879	 lr=0.000100
Epoch[088] Batch [2649]/[3760]	Speed: 70.340696 samples/sec	 accuracy=70.395047	 loss=1.199406	 lr=0.000100
Epoch[088] Batch [2699]/[3760]	Speed: 69.618834 samples/sec	 accuracy=70.388889	 loss=1.199648	 lr=0.000100
Epoch[088] Batch [2749]/[3760]	Speed: 70.055784 samples/sec	 accuracy=70.413636	 loss=1.198252	 lr=0.000100
Epoch[088] Batch [2799]/[3760]	Speed: 69.577881 samples/sec	 accuracy=70.417969	 loss=1.198342	 lr=0.000100
Epoch[088] Batch [2849]/[3760]	Speed: 70.538494 samples/sec	 accuracy=70.424342	 loss=1.197853	 lr=0.000100
Epoch[088] Batch [2899]/[3760]	Speed: 66.077996 samples/sec	 accuracy=70.445582	 loss=1.196856	 lr=0.000100
Epoch[088] Batch [2949]/[3760]	Speed: 72.842687 samples/sec	 accuracy=70.461335	 loss=1.196077	 lr=0.000100
Epoch[088] Batch [2999]/[3760]	Speed: 68.781546 samples/sec	 accuracy=70.449479	 loss=1.195896	 lr=0.000100
Epoch[088] Batch [3049]/[3760]	Speed: 69.759285 samples/sec	 accuracy=70.458504	 loss=1.195505	 lr=0.000100
Epoch[088] Batch [3099]/[3760]	Speed: 70.260246 samples/sec	 accuracy=70.470766	 loss=1.194854	 lr=0.000100
Epoch[088] Batch [3149]/[3760]	Speed: 69.110109 samples/sec	 accuracy=70.470238	 loss=1.194800	 lr=0.000100
Epoch[088] Batch [3199]/[3760]	Speed: 70.280364 samples/sec	 accuracy=70.470215	 loss=1.195013	 lr=0.000100
Epoch[088] Batch [3249]/[3760]	Speed: 70.124403 samples/sec	 accuracy=70.446635	 loss=1.195665	 lr=0.000100
Epoch[088] Batch [3299]/[3760]	Speed: 69.116803 samples/sec	 accuracy=70.438920	 loss=1.195961	 lr=0.000100
Epoch[088] Batch [3349]/[3760]	Speed: 69.365311 samples/sec	 accuracy=70.443097	 loss=1.195941	 lr=0.000100
Epoch[088] Batch [3399]/[3760]	Speed: 70.127417 samples/sec	 accuracy=70.423254	 loss=1.196474	 lr=0.000100
Epoch[088] Batch [3449]/[3760]	Speed: 69.403160 samples/sec	 accuracy=70.431159	 loss=1.196243	 lr=0.000100
Epoch[088] Batch [3499]/[3760]	Speed: 70.186891 samples/sec	 accuracy=70.443750	 loss=1.195643	 lr=0.000100
Epoch[088] Batch [3549]/[3760]	Speed: 70.378886 samples/sec	 accuracy=70.444542	 loss=1.195930	 lr=0.000100
Epoch[088] Batch [3599]/[3760]	Speed: 69.568055 samples/sec	 accuracy=70.436632	 loss=1.196046	 lr=0.000100
Epoch[088] Batch [3649]/[3760]	Speed: 69.539459 samples/sec	 accuracy=70.427226	 loss=1.196525	 lr=0.000100
Epoch[088] Batch [3699]/[3760]	Speed: 70.985669 samples/sec	 accuracy=70.445101	 loss=1.195516	 lr=0.000100
Epoch[088] Batch [3749]/[3760]	Speed: 76.623829 samples/sec	 accuracy=70.443750	 loss=1.195427	 lr=0.000100
Batch [0049]/[0303]: acc-top1=62.906250 acc-top5=84.500000
Batch [0099]/[0303]: acc-top1=62.859375 acc-top5=84.328125
Batch [0149]/[0303]: acc-top1=63.895833 acc-top5=84.864583
Batch [0199]/[0303]: acc-top1=64.406250 acc-top5=85.210938
Batch [0249]/[0303]: acc-top1=64.612500 acc-top5=85.381250
Batch [0299]/[0303]: acc-top1=64.380208 acc-top5=85.177083
[Epoch 088] training: accuracy=70.443401	 loss=1.195414
[Epoch 088] speed: 69 samples/sec	time cost: 3752.794502
[Epoch 088] validation: acc-top1=64.366749 acc-top5=85.153672 loss=1.552372
Epoch[089] Batch [0049]/[3760]	Speed: 47.090358 samples/sec	 accuracy=70.718750	 loss=1.156508	 lr=0.000100
Epoch[089] Batch [0099]/[3760]	Speed: 70.726559 samples/sec	 accuracy=71.687500	 loss=1.140772	 lr=0.000100
Epoch[089] Batch [0149]/[3760]	Speed: 69.694687 samples/sec	 accuracy=71.489583	 loss=1.144592	 lr=0.000100
Epoch[089] Batch [0199]/[3760]	Speed: 69.861079 samples/sec	 accuracy=71.148438	 loss=1.163278	 lr=0.000100
Epoch[089] Batch [0249]/[3760]	Speed: 69.697274 samples/sec	 accuracy=70.706250	 loss=1.182539	 lr=0.000100
Epoch[089] Batch [0299]/[3760]	Speed: 69.580309 samples/sec	 accuracy=70.718750	 loss=1.178433	 lr=0.000100
Epoch[089] Batch [0349]/[3760]	Speed: 71.322254 samples/sec	 accuracy=70.718750	 loss=1.178558	 lr=0.000100
Epoch[089] Batch [0399]/[3760]	Speed: 69.759326 samples/sec	 accuracy=70.644531	 loss=1.178509	 lr=0.000100
Epoch[089] Batch [0449]/[3760]	Speed: 69.509980 samples/sec	 accuracy=70.638889	 loss=1.180842	 lr=0.000100
Epoch[089] Batch [0499]/[3760]	Speed: 69.411719 samples/sec	 accuracy=70.750000	 loss=1.176998	 lr=0.000100
Epoch[089] Batch [0549]/[3760]	Speed: 70.347000 samples/sec	 accuracy=70.556818	 loss=1.184554	 lr=0.000100
Epoch[089] Batch [0599]/[3760]	Speed: 70.341780 samples/sec	 accuracy=70.669271	 loss=1.182790	 lr=0.000100
Epoch[089] Batch [0649]/[3760]	Speed: 69.697603 samples/sec	 accuracy=70.670673	 loss=1.182287	 lr=0.000100
Epoch[089] Batch [0699]/[3760]	Speed: 69.022060 samples/sec	 accuracy=70.689732	 loss=1.181731	 lr=0.000100
Epoch[089] Batch [0749]/[3760]	Speed: 69.757917 samples/sec	 accuracy=70.631250	 loss=1.181587	 lr=0.000100
Epoch[089] Batch [0799]/[3760]	Speed: 69.911582 samples/sec	 accuracy=70.617188	 loss=1.183578	 lr=0.000100
Epoch[089] Batch [0849]/[3760]	Speed: 70.966763 samples/sec	 accuracy=70.623162	 loss=1.182126	 lr=0.000100
Epoch[089] Batch [0899]/[3760]	Speed: 67.076107 samples/sec	 accuracy=70.690972	 loss=1.181588	 lr=0.000100
Epoch[089] Batch [0949]/[3760]	Speed: 69.072440 samples/sec	 accuracy=70.669408	 loss=1.181994	 lr=0.000100
Epoch[089] Batch [0999]/[3760]	Speed: 72.717636 samples/sec	 accuracy=70.715625	 loss=1.181998	 lr=0.000100
Epoch[089] Batch [1049]/[3760]	Speed: 67.531065 samples/sec	 accuracy=70.666667	 loss=1.183758	 lr=0.000100
Epoch[089] Batch [1099]/[3760]	Speed: 70.289476 samples/sec	 accuracy=70.674716	 loss=1.184188	 lr=0.000100
Epoch[089] Batch [1149]/[3760]	Speed: 70.711614 samples/sec	 accuracy=70.698370	 loss=1.184561	 lr=0.000100
Epoch[089] Batch [1199]/[3760]	Speed: 69.660122 samples/sec	 accuracy=70.699219	 loss=1.184205	 lr=0.000100
Epoch[089] Batch [1249]/[3760]	Speed: 70.003729 samples/sec	 accuracy=70.645000	 loss=1.187559	 lr=0.000100
Epoch[089] Batch [1299]/[3760]	Speed: 68.902256 samples/sec	 accuracy=70.640625	 loss=1.186503	 lr=0.000100
Epoch[089] Batch [1349]/[3760]	Speed: 71.737263 samples/sec	 accuracy=70.681713	 loss=1.184369	 lr=0.000100
Epoch[089] Batch [1399]/[3760]	Speed: 69.092999 samples/sec	 accuracy=70.674107	 loss=1.184778	 lr=0.000100
Epoch[089] Batch [1449]/[3760]	Speed: 69.798141 samples/sec	 accuracy=70.647629	 loss=1.186156	 lr=0.000100
Epoch[089] Batch [1499]/[3760]	Speed: 69.205422 samples/sec	 accuracy=70.643750	 loss=1.186366	 lr=0.000100
Epoch[089] Batch [1549]/[3760]	Speed: 70.248665 samples/sec	 accuracy=70.666331	 loss=1.185139	 lr=0.000100
Epoch[089] Batch [1599]/[3760]	Speed: 67.066427 samples/sec	 accuracy=70.709961	 loss=1.184252	 lr=0.000100
Epoch[089] Batch [1649]/[3760]	Speed: 71.165570 samples/sec	 accuracy=70.703598	 loss=1.184879	 lr=0.000100
Epoch[089] Batch [1699]/[3760]	Speed: 71.331168 samples/sec	 accuracy=70.720588	 loss=1.183980	 lr=0.000100
Epoch[089] Batch [1749]/[3760]	Speed: 68.934141 samples/sec	 accuracy=70.711607	 loss=1.184125	 lr=0.000100
Epoch[089] Batch [1799]/[3760]	Speed: 68.034941 samples/sec	 accuracy=70.678819	 loss=1.185300	 lr=0.000100
Epoch[089] Batch [1849]/[3760]	Speed: 71.558210 samples/sec	 accuracy=70.673142	 loss=1.184940	 lr=0.000100
Epoch[089] Batch [1899]/[3760]	Speed: 69.441499 samples/sec	 accuracy=70.646382	 loss=1.186095	 lr=0.000100
Epoch[089] Batch [1949]/[3760]	Speed: 68.865490 samples/sec	 accuracy=70.662660	 loss=1.186409	 lr=0.000100
Epoch[089] Batch [1999]/[3760]	Speed: 69.493824 samples/sec	 accuracy=70.648438	 loss=1.186569	 lr=0.000100
Epoch[089] Batch [2049]/[3760]	Speed: 70.658553 samples/sec	 accuracy=70.652439	 loss=1.186783	 lr=0.000100
Epoch[089] Batch [2099]/[3760]	Speed: 69.855282 samples/sec	 accuracy=70.618304	 loss=1.186790	 lr=0.000100
Epoch[089] Batch [2149]/[3760]	Speed: 69.524204 samples/sec	 accuracy=70.629360	 loss=1.186102	 lr=0.000100
Epoch[089] Batch [2199]/[3760]	Speed: 69.972328 samples/sec	 accuracy=70.648438	 loss=1.185501	 lr=0.000100
Epoch[089] Batch [2249]/[3760]	Speed: 68.646830 samples/sec	 accuracy=70.629861	 loss=1.186213	 lr=0.000100
Epoch[089] Batch [2299]/[3760]	Speed: 69.368093 samples/sec	 accuracy=70.615489	 loss=1.186804	 lr=0.000100
Epoch[089] Batch [2349]/[3760]	Speed: 70.060856 samples/sec	 accuracy=70.625000	 loss=1.186748	 lr=0.000100
Epoch[089] Batch [2399]/[3760]	Speed: 70.709394 samples/sec	 accuracy=70.612630	 loss=1.187142	 lr=0.000100
Epoch[089] Batch [2449]/[3760]	Speed: 67.813215 samples/sec	 accuracy=70.618622	 loss=1.186608	 lr=0.000100
Epoch[089] Batch [2499]/[3760]	Speed: 71.069923 samples/sec	 accuracy=70.626875	 loss=1.186724	 lr=0.000100
Epoch[089] Batch [2549]/[3760]	Speed: 69.120956 samples/sec	 accuracy=70.593750	 loss=1.187949	 lr=0.000100
Epoch[089] Batch [2599]/[3760]	Speed: 69.764967 samples/sec	 accuracy=70.576923	 loss=1.188498	 lr=0.000100
Epoch[089] Batch [2649]/[3760]	Speed: 70.596358 samples/sec	 accuracy=70.599057	 loss=1.187418	 lr=0.000100
Epoch[089] Batch [2699]/[3760]	Speed: 70.679152 samples/sec	 accuracy=70.612269	 loss=1.187311	 lr=0.000100
Epoch[089] Batch [2749]/[3760]	Speed: 69.480288 samples/sec	 accuracy=70.618182	 loss=1.187340	 lr=0.000100
Epoch[089] Batch [2799]/[3760]	Speed: 69.570247 samples/sec	 accuracy=70.606585	 loss=1.187344	 lr=0.000100
Epoch[089] Batch [2849]/[3760]	Speed: 67.914805 samples/sec	 accuracy=70.607456	 loss=1.187600	 lr=0.000100
Epoch[089] Batch [2899]/[3760]	Speed: 70.500972 samples/sec	 accuracy=70.604526	 loss=1.188323	 lr=0.000100
Epoch[089] Batch [2949]/[3760]	Speed: 68.982122 samples/sec	 accuracy=70.593750	 loss=1.188502	 lr=0.000100
Epoch[089] Batch [2999]/[3760]	Speed: 70.615343 samples/sec	 accuracy=70.596875	 loss=1.188098	 lr=0.000100
Epoch[089] Batch [3049]/[3760]	Speed: 69.988474 samples/sec	 accuracy=70.581455	 loss=1.188734	 lr=0.000100
Epoch[089] Batch [3099]/[3760]	Speed: 67.967767 samples/sec	 accuracy=70.575101	 loss=1.189584	 lr=0.000100
Epoch[089] Batch [3149]/[3760]	Speed: 70.693306 samples/sec	 accuracy=70.565972	 loss=1.189673	 lr=0.000100
Epoch[089] Batch [3199]/[3760]	Speed: 70.870417 samples/sec	 accuracy=70.557129	 loss=1.189768	 lr=0.000100
Epoch[089] Batch [3249]/[3760]	Speed: 68.999492 samples/sec	 accuracy=70.579327	 loss=1.189105	 lr=0.000100
Epoch[089] Batch [3299]/[3760]	Speed: 69.849900 samples/sec	 accuracy=70.589962	 loss=1.189214	 lr=0.000100
Epoch[089] Batch [3349]/[3760]	Speed: 68.685363 samples/sec	 accuracy=70.600746	 loss=1.189123	 lr=0.000100
Epoch[089] Batch [3399]/[3760]	Speed: 70.798709 samples/sec	 accuracy=70.611213	 loss=1.188649	 lr=0.000100
Epoch[089] Batch [3449]/[3760]	Speed: 70.205664 samples/sec	 accuracy=70.605072	 loss=1.188965	 lr=0.000100
Epoch[089] Batch [3499]/[3760]	Speed: 68.046067 samples/sec	 accuracy=70.595982	 loss=1.189245	 lr=0.000100
Epoch[089] Batch [3549]/[3760]	Speed: 70.896695 samples/sec	 accuracy=70.580546	 loss=1.189751	 lr=0.000100
Epoch[089] Batch [3599]/[3760]	Speed: 69.849979 samples/sec	 accuracy=70.571181	 loss=1.189838	 lr=0.000100
Epoch[089] Batch [3649]/[3760]	Speed: 69.937573 samples/sec	 accuracy=70.556935	 loss=1.190439	 lr=0.000100
Epoch[089] Batch [3699]/[3760]	Speed: 69.593437 samples/sec	 accuracy=70.559544	 loss=1.190504	 lr=0.000100
Epoch[089] Batch [3749]/[3760]	Speed: 77.075636 samples/sec	 accuracy=70.562083	 loss=1.190137	 lr=0.000100
Batch [0049]/[0303]: acc-top1=62.937500 acc-top5=83.937500
Batch [0099]/[0303]: acc-top1=62.875000 acc-top5=84.171875
Batch [0149]/[0303]: acc-top1=63.947917 acc-top5=84.531250
Batch [0199]/[0303]: acc-top1=64.328125 acc-top5=84.921875
Batch [0249]/[0303]: acc-top1=64.593750 acc-top5=85.156250
Batch [0299]/[0303]: acc-top1=64.421875 acc-top5=84.968750
[Epoch 089] training: accuracy=70.557680	 loss=1.190479
[Epoch 089] speed: 69 samples/sec	time cost: 3751.070344
[Epoch 089] validation: acc-top1=64.387376 acc-top5=84.978342 loss=1.561762
Epoch[090] Batch [0049]/[3759]	Speed: 49.449888 samples/sec	 accuracy=70.625000	 loss=1.183235	 lr=0.000100
Epoch[090] Batch [0099]/[3759]	Speed: 67.485512 samples/sec	 accuracy=70.109375	 loss=1.190769	 lr=0.000100
Epoch[090] Batch [0149]/[3759]	Speed: 70.444743 samples/sec	 accuracy=70.218750	 loss=1.194773	 lr=0.000100
Epoch[090] Batch [0199]/[3759]	Speed: 70.332929 samples/sec	 accuracy=70.320312	 loss=1.194426	 lr=0.000100
Epoch[090] Batch [0249]/[3759]	Speed: 68.484987 samples/sec	 accuracy=70.137500	 loss=1.199380	 lr=0.000100
Epoch[090] Batch [0299]/[3759]	Speed: 69.670498 samples/sec	 accuracy=70.296875	 loss=1.192009	 lr=0.000100
Epoch[090] Batch [0349]/[3759]	Speed: 69.701271 samples/sec	 accuracy=70.375000	 loss=1.191517	 lr=0.000100
Epoch[090] Batch [0399]/[3759]	Speed: 69.778165 samples/sec	 accuracy=70.273438	 loss=1.193749	 lr=0.000100
Epoch[090] Batch [0449]/[3759]	Speed: 69.525329 samples/sec	 accuracy=70.250000	 loss=1.195682	 lr=0.000100
Epoch[090] Batch [0499]/[3759]	Speed: 69.329294 samples/sec	 accuracy=70.306250	 loss=1.194063	 lr=0.000100
Epoch[090] Batch [0549]/[3759]	Speed: 69.996888 samples/sec	 accuracy=70.318182	 loss=1.193221	 lr=0.000100
Epoch[090] Batch [0599]/[3759]	Speed: 69.622975 samples/sec	 accuracy=70.414062	 loss=1.189704	 lr=0.000100
Epoch[090] Batch [0649]/[3759]	Speed: 69.575528 samples/sec	 accuracy=70.399038	 loss=1.193455	 lr=0.000100
Epoch[090] Batch [0699]/[3759]	Speed: 69.790954 samples/sec	 accuracy=70.379464	 loss=1.195096	 lr=0.000100
Epoch[090] Batch [0749]/[3759]	Speed: 70.908483 samples/sec	 accuracy=70.437500	 loss=1.192588	 lr=0.000100
Epoch[090] Batch [0799]/[3759]	Speed: 69.184062 samples/sec	 accuracy=70.462891	 loss=1.190523	 lr=0.000100
Epoch[090] Batch [0849]/[3759]	Speed: 68.208056 samples/sec	 accuracy=70.529412	 loss=1.188035	 lr=0.000100
Epoch[090] Batch [0899]/[3759]	Speed: 71.715194 samples/sec	 accuracy=70.588542	 loss=1.184543	 lr=0.000100
Epoch[090] Batch [0949]/[3759]	Speed: 69.471971 samples/sec	 accuracy=70.526316	 loss=1.184491	 lr=0.000100
Epoch[090] Batch [0999]/[3759]	Speed: 70.530025 samples/sec	 accuracy=70.576562	 loss=1.182574	 lr=0.000100
Epoch[090] Batch [1049]/[3759]	Speed: 69.597027 samples/sec	 accuracy=70.549107	 loss=1.183059	 lr=0.000100
Epoch[090] Batch [1099]/[3759]	Speed: 69.169334 samples/sec	 accuracy=70.468750	 loss=1.185712	 lr=0.000100
Epoch[090] Batch [1149]/[3759]	Speed: 69.278882 samples/sec	 accuracy=70.478261	 loss=1.185745	 lr=0.000100
Epoch[090] Batch [1199]/[3759]	Speed: 70.378262 samples/sec	 accuracy=70.497396	 loss=1.185247	 lr=0.000100
Epoch[090] Batch [1249]/[3759]	Speed: 69.709912 samples/sec	 accuracy=70.465000	 loss=1.187286	 lr=0.000100
Epoch[090] Batch [1299]/[3759]	Speed: 67.645257 samples/sec	 accuracy=70.449519	 loss=1.187999	 lr=0.000100
Epoch[090] Batch [1349]/[3759]	Speed: 69.007895 samples/sec	 accuracy=70.407407	 loss=1.189170	 lr=0.000100
Epoch[090] Batch [1399]/[3759]	Speed: 72.602944 samples/sec	 accuracy=70.430804	 loss=1.188065	 lr=0.000100
Epoch[090] Batch [1449]/[3759]	Speed: 69.358708 samples/sec	 accuracy=70.470905	 loss=1.187030	 lr=0.000100
Epoch[090] Batch [1499]/[3759]	Speed: 70.924112 samples/sec	 accuracy=70.396875	 loss=1.189700	 lr=0.000100
Epoch[090] Batch [1549]/[3759]	Speed: 69.885623 samples/sec	 accuracy=70.429435	 loss=1.188616	 lr=0.000100
Epoch[090] Batch [1599]/[3759]	Speed: 69.628543 samples/sec	 accuracy=70.409180	 loss=1.187712	 lr=0.000100
Epoch[090] Batch [1649]/[3759]	Speed: 68.844115 samples/sec	 accuracy=70.367424	 loss=1.189717	 lr=0.000100
Epoch[090] Batch [1699]/[3759]	Speed: 69.598861 samples/sec	 accuracy=70.376838	 loss=1.190018	 lr=0.000100
Epoch[090] Batch [1749]/[3759]	Speed: 70.710606 samples/sec	 accuracy=70.388393	 loss=1.189982	 lr=0.000100
Epoch[090] Batch [1799]/[3759]	Speed: 66.548410 samples/sec	 accuracy=70.414062	 loss=1.188957	 lr=0.000100
Epoch[090] Batch [1849]/[3759]	Speed: 72.706331 samples/sec	 accuracy=70.415541	 loss=1.189359	 lr=0.000100
Epoch[090] Batch [1899]/[3759]	Speed: 69.894513 samples/sec	 accuracy=70.416118	 loss=1.189665	 lr=0.000100
Epoch[090] Batch [1949]/[3759]	Speed: 70.681885 samples/sec	 accuracy=70.403045	 loss=1.190429	 lr=0.000100
Epoch[090] Batch [1999]/[3759]	Speed: 68.075677 samples/sec	 accuracy=70.413281	 loss=1.191059	 lr=0.000100
Epoch[090] Batch [2049]/[3759]	Speed: 69.296460 samples/sec	 accuracy=70.438262	 loss=1.189193	 lr=0.000100
Epoch[090] Batch [2099]/[3759]	Speed: 70.932979 samples/sec	 accuracy=70.442708	 loss=1.189435	 lr=0.000100
Epoch[090] Batch [2149]/[3759]	Speed: 70.222863 samples/sec	 accuracy=70.400436	 loss=1.190997	 lr=0.000100
Epoch[090] Batch [2199]/[3759]	Speed: 68.040023 samples/sec	 accuracy=70.390625	 loss=1.192010	 lr=0.000100
Epoch[090] Batch [2249]/[3759]	Speed: 68.050667 samples/sec	 accuracy=70.422222	 loss=1.190650	 lr=0.000100
Epoch[090] Batch [2299]/[3759]	Speed: 72.532568 samples/sec	 accuracy=70.407609	 loss=1.191319	 lr=0.000100
Epoch[090] Batch [2349]/[3759]	Speed: 69.906777 samples/sec	 accuracy=70.425532	 loss=1.190233	 lr=0.000100
Epoch[090] Batch [2399]/[3759]	Speed: 69.234287 samples/sec	 accuracy=70.415365	 loss=1.189955	 lr=0.000100
Epoch[090] Batch [2449]/[3759]	Speed: 70.318215 samples/sec	 accuracy=70.422832	 loss=1.189453	 lr=0.000100
Epoch[090] Batch [2499]/[3759]	Speed: 68.575183 samples/sec	 accuracy=70.436250	 loss=1.189514	 lr=0.000100
Epoch[090] Batch [2549]/[3759]	Speed: 70.868218 samples/sec	 accuracy=70.418505	 loss=1.190614	 lr=0.000100
Epoch[090] Batch [2599]/[3759]	Speed: 69.725556 samples/sec	 accuracy=70.430889	 loss=1.190139	 lr=0.000100
Epoch[090] Batch [2649]/[3759]	Speed: 69.908384 samples/sec	 accuracy=70.409788	 loss=1.190318	 lr=0.000100
Epoch[090] Batch [2699]/[3759]	Speed: 68.978216 samples/sec	 accuracy=70.416088	 loss=1.190163	 lr=0.000100
Epoch[090] Batch [2749]/[3759]	Speed: 70.877138 samples/sec	 accuracy=70.414205	 loss=1.190338	 lr=0.000100
Epoch[090] Batch [2799]/[3759]	Speed: 69.439973 samples/sec	 accuracy=70.416853	 loss=1.189859	 lr=0.000100
Epoch[090] Batch [2849]/[3759]	Speed: 67.144754 samples/sec	 accuracy=70.416667	 loss=1.190294	 lr=0.000100
Epoch[090] Batch [2899]/[3759]	Speed: 71.755419 samples/sec	 accuracy=70.418642	 loss=1.190010	 lr=0.000100
Epoch[090] Batch [2949]/[3759]	Speed: 70.443915 samples/sec	 accuracy=70.408898	 loss=1.190486	 lr=0.000100
Epoch[090] Batch [2999]/[3759]	Speed: 69.716858 samples/sec	 accuracy=70.414062	 loss=1.189818	 lr=0.000100
Epoch[090] Batch [3049]/[3759]	Speed: 68.204200 samples/sec	 accuracy=70.413422	 loss=1.190106	 lr=0.000100
Epoch[090] Batch [3099]/[3759]	Speed: 70.345833 samples/sec	 accuracy=70.420363	 loss=1.190331	 lr=0.000100
Epoch[090] Batch [3149]/[3759]	Speed: 69.316593 samples/sec	 accuracy=70.427579	 loss=1.190143	 lr=0.000100
Epoch[090] Batch [3199]/[3759]	Speed: 69.206144 samples/sec	 accuracy=70.403320	 loss=1.190884	 lr=0.000100
Epoch[090] Batch [3249]/[3759]	Speed: 70.758433 samples/sec	 accuracy=70.408173	 loss=1.191000	 lr=0.000100
Epoch[090] Batch [3299]/[3759]	Speed: 69.516843 samples/sec	 accuracy=70.406723	 loss=1.190731	 lr=0.000100
Epoch[090] Batch [3349]/[3759]	Speed: 69.024951 samples/sec	 accuracy=70.420709	 loss=1.190249	 lr=0.000100
Epoch[090] Batch [3399]/[3759]	Speed: 69.119972 samples/sec	 accuracy=70.434283	 loss=1.189393	 lr=0.000100
Epoch[090] Batch [3449]/[3759]	Speed: 69.263122 samples/sec	 accuracy=70.438859	 loss=1.189526	 lr=0.000100
Epoch[090] Batch [3499]/[3759]	Speed: 69.647517 samples/sec	 accuracy=70.422321	 loss=1.190173	 lr=0.000100
Epoch[090] Batch [3549]/[3759]	Speed: 69.997902 samples/sec	 accuracy=70.425176	 loss=1.190116	 lr=0.000100
Epoch[090] Batch [3599]/[3759]	Speed: 67.876102 samples/sec	 accuracy=70.424913	 loss=1.189654	 lr=0.000100
Epoch[090] Batch [3649]/[3759]	Speed: 69.565780 samples/sec	 accuracy=70.443493	 loss=1.189071	 lr=0.000100
Epoch[090] Batch [3699]/[3759]	Speed: 70.861573 samples/sec	 accuracy=70.438767	 loss=1.189234	 lr=0.000100
Epoch[090] Batch [3749]/[3759]	Speed: 76.880564 samples/sec	 accuracy=70.444167	 loss=1.189327	 lr=0.000100
Batch [0049]/[0303]: acc-top1=62.750000 acc-top5=84.250000
Batch [0099]/[0303]: acc-top1=62.812500 acc-top5=84.250000
Batch [0149]/[0303]: acc-top1=63.895833 acc-top5=84.687500
Batch [0199]/[0303]: acc-top1=64.375000 acc-top5=85.101562
Batch [0249]/[0303]: acc-top1=64.606250 acc-top5=85.293750
Batch [0299]/[0303]: acc-top1=64.432292 acc-top5=85.125000
[Epoch 090] training: accuracy=70.445514	 loss=1.189154
[Epoch 090] speed: 69 samples/sec	time cost: 3750.459925
[Epoch 090] validation: acc-top1=64.397690 acc-top5=85.127888 loss=1.553392
Epoch[091] Batch [0049]/[3760]	Speed: 48.375856 samples/sec	 accuracy=71.812500	 loss=1.135260	 lr=0.000100
Epoch[091] Batch [0099]/[3760]	Speed: 69.549998 samples/sec	 accuracy=71.234375	 loss=1.154775	 lr=0.000100
Epoch[091] Batch [0149]/[3760]	Speed: 68.999712 samples/sec	 accuracy=71.135417	 loss=1.164828	 lr=0.000100
Epoch[091] Batch [0199]/[3760]	Speed: 69.931655 samples/sec	 accuracy=70.789062	 loss=1.177342	 lr=0.000100
Epoch[091] Batch [0249]/[3760]	Speed: 70.343916 samples/sec	 accuracy=70.656250	 loss=1.171113	 lr=0.000100
Epoch[091] Batch [0299]/[3760]	Speed: 69.686904 samples/sec	 accuracy=70.583333	 loss=1.168681	 lr=0.000100
Epoch[091] Batch [0349]/[3760]	Speed: 70.667166 samples/sec	 accuracy=70.651786	 loss=1.167913	 lr=0.000100
Epoch[091] Batch [0399]/[3760]	Speed: 69.999301 samples/sec	 accuracy=70.726562	 loss=1.163644	 lr=0.000100
Epoch[091] Batch [0449]/[3760]	Speed: 69.003067 samples/sec	 accuracy=70.565972	 loss=1.172076	 lr=0.000100
Epoch[091] Batch [0499]/[3760]	Speed: 70.273602 samples/sec	 accuracy=70.534375	 loss=1.172817	 lr=0.000100
Epoch[091] Batch [0549]/[3760]	Speed: 69.483846 samples/sec	 accuracy=70.539773	 loss=1.174006	 lr=0.000100
Epoch[091] Batch [0599]/[3760]	Speed: 70.372772 samples/sec	 accuracy=70.500000	 loss=1.177812	 lr=0.000100
Epoch[091] Batch [0649]/[3760]	Speed: 68.546751 samples/sec	 accuracy=70.574519	 loss=1.176550	 lr=0.000100
Epoch[091] Batch [0699]/[3760]	Speed: 68.996816 samples/sec	 accuracy=70.662946	 loss=1.174072	 lr=0.000100
Epoch[091] Batch [0749]/[3760]	Speed: 70.698479 samples/sec	 accuracy=70.581250	 loss=1.177695	 lr=0.000100
Epoch[091] Batch [0799]/[3760]	Speed: 69.178754 samples/sec	 accuracy=70.585938	 loss=1.177082	 lr=0.000100
Epoch[091] Batch [0849]/[3760]	Speed: 69.835056 samples/sec	 accuracy=70.470588	 loss=1.179591	 lr=0.000100
Epoch[091] Batch [0899]/[3760]	Speed: 69.199938 samples/sec	 accuracy=70.487847	 loss=1.180609	 lr=0.000100
Epoch[091] Batch [0949]/[3760]	Speed: 70.007129 samples/sec	 accuracy=70.457237	 loss=1.182294	 lr=0.000100
Epoch[091] Batch [0999]/[3760]	Speed: 68.961093 samples/sec	 accuracy=70.409375	 loss=1.183257	 lr=0.000100
Epoch[091] Batch [1049]/[3760]	Speed: 71.204223 samples/sec	 accuracy=70.398810	 loss=1.184689	 lr=0.000100
Epoch[091] Batch [1099]/[3760]	Speed: 70.365097 samples/sec	 accuracy=70.447443	 loss=1.184147	 lr=0.000100
Epoch[091] Batch [1149]/[3760]	Speed: 69.377466 samples/sec	 accuracy=70.423913	 loss=1.185323	 lr=0.000100
Epoch[091] Batch [1199]/[3760]	Speed: 69.691449 samples/sec	 accuracy=70.425781	 loss=1.185929	 lr=0.000100
Epoch[091] Batch [1249]/[3760]	Speed: 70.407937 samples/sec	 accuracy=70.400000	 loss=1.186423	 lr=0.000100
Epoch[091] Batch [1299]/[3760]	Speed: 70.349366 samples/sec	 accuracy=70.389423	 loss=1.186078	 lr=0.000100
Epoch[091] Batch [1349]/[3760]	Speed: 69.743241 samples/sec	 accuracy=70.451389	 loss=1.185015	 lr=0.000100
Epoch[091] Batch [1399]/[3760]	Speed: 69.041942 samples/sec	 accuracy=70.464286	 loss=1.184874	 lr=0.000100
Epoch[091] Batch [1449]/[3760]	Speed: 69.636242 samples/sec	 accuracy=70.467672	 loss=1.184550	 lr=0.000100
Epoch[091] Batch [1499]/[3760]	Speed: 69.517103 samples/sec	 accuracy=70.461458	 loss=1.184276	 lr=0.000100
Epoch[091] Batch [1549]/[3760]	Speed: 70.679097 samples/sec	 accuracy=70.512097	 loss=1.182178	 lr=0.000100
Epoch[091] Batch [1599]/[3760]	Speed: 69.592777 samples/sec	 accuracy=70.514648	 loss=1.181674	 lr=0.000100
Epoch[091] Batch [1649]/[3760]	Speed: 69.636934 samples/sec	 accuracy=70.514205	 loss=1.182291	 lr=0.000100
Epoch[091] Batch [1699]/[3760]	Speed: 71.063065 samples/sec	 accuracy=70.502757	 loss=1.182708	 lr=0.000100
Epoch[091] Batch [1749]/[3760]	Speed: 69.727398 samples/sec	 accuracy=70.529464	 loss=1.181745	 lr=0.000100
Epoch[091] Batch [1799]/[3760]	Speed: 68.973922 samples/sec	 accuracy=70.538194	 loss=1.180792	 lr=0.000100
Epoch[091] Batch [1849]/[3760]	Speed: 69.420663 samples/sec	 accuracy=70.565034	 loss=1.179278	 lr=0.000100
Epoch[091] Batch [1899]/[3760]	Speed: 69.621312 samples/sec	 accuracy=70.546053	 loss=1.181554	 lr=0.000100
Epoch[091] Batch [1949]/[3760]	Speed: 70.133815 samples/sec	 accuracy=70.580128	 loss=1.180772	 lr=0.000100
Epoch[091] Batch [1999]/[3760]	Speed: 68.758038 samples/sec	 accuracy=70.597656	 loss=1.180421	 lr=0.000100
Epoch[091] Batch [2049]/[3760]	Speed: 70.606002 samples/sec	 accuracy=70.581555	 loss=1.179815	 lr=0.000100
Epoch[091] Batch [2099]/[3760]	Speed: 68.768470 samples/sec	 accuracy=70.574405	 loss=1.180233	 lr=0.000100
Epoch[091] Batch [2149]/[3760]	Speed: 70.729341 samples/sec	 accuracy=70.569767	 loss=1.180777	 lr=0.000100
Epoch[091] Batch [2199]/[3760]	Speed: 69.499336 samples/sec	 accuracy=70.605114	 loss=1.180263	 lr=0.000100
Epoch[091] Batch [2249]/[3760]	Speed: 70.681684 samples/sec	 accuracy=70.621528	 loss=1.179867	 lr=0.000100
Epoch[091] Batch [2299]/[3760]	Speed: 70.146338 samples/sec	 accuracy=70.611413	 loss=1.181274	 lr=0.000100
Epoch[091] Batch [2349]/[3760]	Speed: 68.468030 samples/sec	 accuracy=70.628989	 loss=1.180848	 lr=0.000100
Epoch[091] Batch [2399]/[3760]	Speed: 71.257695 samples/sec	 accuracy=70.638021	 loss=1.180637	 lr=0.000100
Epoch[091] Batch [2449]/[3760]	Speed: 69.129926 samples/sec	 accuracy=70.610332	 loss=1.181955	 lr=0.000100
Epoch[091] Batch [2499]/[3760]	Speed: 70.001131 samples/sec	 accuracy=70.588750	 loss=1.182460	 lr=0.000100
Epoch[091] Batch [2549]/[3760]	Speed: 69.943320 samples/sec	 accuracy=70.594363	 loss=1.182518	 lr=0.000100
Epoch[091] Batch [2599]/[3760]	Speed: 69.814325 samples/sec	 accuracy=70.585337	 loss=1.182697	 lr=0.000100
Epoch[091] Batch [2649]/[3760]	Speed: 69.521740 samples/sec	 accuracy=70.554245	 loss=1.183721	 lr=0.000100
Epoch[091] Batch [2699]/[3760]	Speed: 69.906590 samples/sec	 accuracy=70.560185	 loss=1.183837	 lr=0.000100
Epoch[091] Batch [2749]/[3760]	Speed: 69.696036 samples/sec	 accuracy=70.572159	 loss=1.184519	 lr=0.000100
Epoch[091] Batch [2799]/[3760]	Speed: 69.903501 samples/sec	 accuracy=70.593192	 loss=1.184085	 lr=0.000100
Epoch[091] Batch [2849]/[3760]	Speed: 70.492034 samples/sec	 accuracy=70.582237	 loss=1.184890	 lr=0.000100
Epoch[091] Batch [2899]/[3760]	Speed: 66.762234 samples/sec	 accuracy=70.581358	 loss=1.185168	 lr=0.000100
Epoch[091] Batch [2949]/[3760]	Speed: 71.958709 samples/sec	 accuracy=70.587924	 loss=1.185210	 lr=0.000100
Epoch[091] Batch [2999]/[3760]	Speed: 70.344677 samples/sec	 accuracy=70.604687	 loss=1.184668	 lr=0.000100
Epoch[091] Batch [3049]/[3760]	Speed: 69.172602 samples/sec	 accuracy=70.592213	 loss=1.184492	 lr=0.000100
Epoch[091] Batch [3099]/[3760]	Speed: 69.513594 samples/sec	 accuracy=70.583165	 loss=1.184965	 lr=0.000100
Epoch[091] Batch [3149]/[3760]	Speed: 70.511143 samples/sec	 accuracy=70.576389	 loss=1.185266	 lr=0.000100
Epoch[091] Batch [3199]/[3760]	Speed: 69.817105 samples/sec	 accuracy=70.585938	 loss=1.185284	 lr=0.000100
Epoch[091] Batch [3249]/[3760]	Speed: 70.145897 samples/sec	 accuracy=70.590385	 loss=1.185338	 lr=0.000100
Epoch[091] Batch [3299]/[3760]	Speed: 68.849574 samples/sec	 accuracy=70.588068	 loss=1.185562	 lr=0.000100
Epoch[091] Batch [3349]/[3760]	Speed: 69.952843 samples/sec	 accuracy=70.592351	 loss=1.185696	 lr=0.000100
Epoch[091] Batch [3399]/[3760]	Speed: 69.614640 samples/sec	 accuracy=70.584559	 loss=1.186158	 lr=0.000100
Epoch[091] Batch [3449]/[3760]	Speed: 70.208706 samples/sec	 accuracy=70.581522	 loss=1.186039	 lr=0.000100
Epoch[091] Batch [3499]/[3760]	Speed: 70.574935 samples/sec	 accuracy=70.560714	 loss=1.186608	 lr=0.000100
Epoch[091] Batch [3549]/[3760]	Speed: 69.916204 samples/sec	 accuracy=70.568662	 loss=1.186385	 lr=0.000100
Epoch[091] Batch [3599]/[3760]	Speed: 68.986317 samples/sec	 accuracy=70.547743	 loss=1.187346	 lr=0.000100
Epoch[091] Batch [3649]/[3760]	Speed: 70.043919 samples/sec	 accuracy=70.553938	 loss=1.187513	 lr=0.000100
Epoch[091] Batch [3699]/[3760]	Speed: 69.382093 samples/sec	 accuracy=70.557010	 loss=1.187321	 lr=0.000100
Epoch[091] Batch [3749]/[3760]	Speed: 76.980607 samples/sec	 accuracy=70.546250	 loss=1.187713	 lr=0.000100
Batch [0049]/[0303]: acc-top1=62.468750 acc-top5=84.218750
Batch [0099]/[0303]: acc-top1=62.640625 acc-top5=84.343750
Batch [0149]/[0303]: acc-top1=63.656250 acc-top5=84.677083
Batch [0199]/[0303]: acc-top1=64.195312 acc-top5=85.031250
Batch [0249]/[0303]: acc-top1=64.500000 acc-top5=85.268750
Batch [0299]/[0303]: acc-top1=64.348958 acc-top5=85.104167
[Epoch 091] training: accuracy=70.552693	 loss=1.187461
[Epoch 091] speed: 69 samples/sec	time cost: 3746.240567
[Epoch 091] validation: acc-top1=64.315182 acc-top5=85.107261 loss=1.552106
Epoch[092] Batch [0049]/[3760]	Speed: 47.629203 samples/sec	 accuracy=70.687500	 loss=1.177189	 lr=0.000100
Epoch[092] Batch [0099]/[3760]	Speed: 69.829726 samples/sec	 accuracy=70.781250	 loss=1.184149	 lr=0.000100
Epoch[092] Batch [0149]/[3760]	Speed: 68.835702 samples/sec	 accuracy=70.708333	 loss=1.182863	 lr=0.000100
Epoch[092] Batch [0199]/[3760]	Speed: 69.742350 samples/sec	 accuracy=70.906250	 loss=1.173085	 lr=0.000100
Epoch[092] Batch [0249]/[3760]	Speed: 70.507350 samples/sec	 accuracy=70.600000	 loss=1.180926	 lr=0.000100
Epoch[092] Batch [0299]/[3760]	Speed: 69.436847 samples/sec	 accuracy=70.739583	 loss=1.179131	 lr=0.000100
Epoch[092] Batch [0349]/[3760]	Speed: 70.303822 samples/sec	 accuracy=70.924107	 loss=1.171156	 lr=0.000100
Epoch[092] Batch [0399]/[3760]	Speed: 68.789574 samples/sec	 accuracy=70.960938	 loss=1.169115	 lr=0.000100
Epoch[092] Batch [0449]/[3760]	Speed: 68.211990 samples/sec	 accuracy=70.881944	 loss=1.174079	 lr=0.000100
Epoch[092] Batch [0499]/[3760]	Speed: 70.374711 samples/sec	 accuracy=70.834375	 loss=1.176008	 lr=0.000100
Epoch[092] Batch [0549]/[3760]	Speed: 70.573107 samples/sec	 accuracy=70.778409	 loss=1.177896	 lr=0.000100
Epoch[092] Batch [0599]/[3760]	Speed: 69.372606 samples/sec	 accuracy=70.666667	 loss=1.181897	 lr=0.000100
Epoch[092] Batch [0649]/[3760]	Speed: 69.516811 samples/sec	 accuracy=70.713942	 loss=1.183288	 lr=0.000100
Epoch[092] Batch [0699]/[3760]	Speed: 68.660583 samples/sec	 accuracy=70.694196	 loss=1.183900	 lr=0.000100
Epoch[092] Batch [0749]/[3760]	Speed: 70.607721 samples/sec	 accuracy=70.764583	 loss=1.183651	 lr=0.000100
Epoch[092] Batch [0799]/[3760]	Speed: 70.308936 samples/sec	 accuracy=70.691406	 loss=1.185596	 lr=0.000100
Epoch[092] Batch [0849]/[3760]	Speed: 69.962627 samples/sec	 accuracy=70.654412	 loss=1.185341	 lr=0.000100
Epoch[092] Batch [0899]/[3760]	Speed: 68.184155 samples/sec	 accuracy=70.642361	 loss=1.184559	 lr=0.000100
Epoch[092] Batch [0949]/[3760]	Speed: 71.286911 samples/sec	 accuracy=70.674342	 loss=1.184319	 lr=0.000100
Epoch[092] Batch [0999]/[3760]	Speed: 70.622575 samples/sec	 accuracy=70.685937	 loss=1.184574	 lr=0.000100
Epoch[092] Batch [1049]/[3760]	Speed: 69.384616 samples/sec	 accuracy=70.705357	 loss=1.184353	 lr=0.000100
Epoch[092] Batch [1099]/[3760]	Speed: 68.189083 samples/sec	 accuracy=70.708807	 loss=1.185773	 lr=0.000100
Epoch[092] Batch [1149]/[3760]	Speed: 70.129455 samples/sec	 accuracy=70.706522	 loss=1.184763	 lr=0.000100
Epoch[092] Batch [1199]/[3760]	Speed: 70.279691 samples/sec	 accuracy=70.720052	 loss=1.185236	 lr=0.000100
Epoch[092] Batch [1249]/[3760]	Speed: 68.882010 samples/sec	 accuracy=70.738750	 loss=1.185023	 lr=0.000100
Epoch[092] Batch [1299]/[3760]	Speed: 69.819520 samples/sec	 accuracy=70.742788	 loss=1.185686	 lr=0.000100
Epoch[092] Batch [1349]/[3760]	Speed: 69.572508 samples/sec	 accuracy=70.752315	 loss=1.184760	 lr=0.000100
Epoch[092] Batch [1399]/[3760]	Speed: 69.633154 samples/sec	 accuracy=70.760045	 loss=1.184806	 lr=0.000100
Epoch[092] Batch [1449]/[3760]	Speed: 68.458871 samples/sec	 accuracy=70.782328	 loss=1.184894	 lr=0.000100
Epoch[092] Batch [1499]/[3760]	Speed: 70.275838 samples/sec	 accuracy=70.787500	 loss=1.184181	 lr=0.000100
Epoch[092] Batch [1549]/[3760]	Speed: 70.525886 samples/sec	 accuracy=70.779234	 loss=1.184888	 lr=0.000100
Epoch[092] Batch [1599]/[3760]	Speed: 70.804063 samples/sec	 accuracy=70.755859	 loss=1.186832	 lr=0.000100
Epoch[092] Batch [1649]/[3760]	Speed: 67.899386 samples/sec	 accuracy=70.756629	 loss=1.186283	 lr=0.000100
Epoch[092] Batch [1699]/[3760]	Speed: 70.520950 samples/sec	 accuracy=70.764706	 loss=1.185242	 lr=0.000100
Epoch[092] Batch [1749]/[3760]	Speed: 70.294791 samples/sec	 accuracy=70.765179	 loss=1.185300	 lr=0.000100
Epoch[092] Batch [1799]/[3760]	Speed: 67.976428 samples/sec	 accuracy=70.799479	 loss=1.184577	 lr=0.000100
Epoch[092] Batch [1849]/[3760]	Speed: 69.327079 samples/sec	 accuracy=70.782939	 loss=1.185715	 lr=0.000100
Epoch[092] Batch [1899]/[3760]	Speed: 68.542258 samples/sec	 accuracy=70.785362	 loss=1.185995	 lr=0.000100
Epoch[092] Batch [1949]/[3760]	Speed: 68.959786 samples/sec	 accuracy=70.794071	 loss=1.185423	 lr=0.000100
Epoch[092] Batch [1999]/[3760]	Speed: 71.870925 samples/sec	 accuracy=70.782813	 loss=1.185684	 lr=0.000100
Epoch[092] Batch [2049]/[3760]	Speed: 67.093462 samples/sec	 accuracy=70.775152	 loss=1.184735	 lr=0.000100
Epoch[092] Batch [2099]/[3760]	Speed: 71.171235 samples/sec	 accuracy=70.760417	 loss=1.185048	 lr=0.000100
Epoch[092] Batch [2149]/[3760]	Speed: 71.243733 samples/sec	 accuracy=70.752907	 loss=1.185631	 lr=0.000100
Epoch[092] Batch [2199]/[3760]	Speed: 67.795609 samples/sec	 accuracy=70.788352	 loss=1.184563	 lr=0.000100
Epoch[092] Batch [2249]/[3760]	Speed: 70.684626 samples/sec	 accuracy=70.789583	 loss=1.185100	 lr=0.000100
Epoch[092] Batch [2299]/[3760]	Speed: 69.171940 samples/sec	 accuracy=70.757473	 loss=1.185872	 lr=0.000100
Epoch[092] Batch [2349]/[3760]	Speed: 69.321990 samples/sec	 accuracy=70.754654	 loss=1.186611	 lr=0.000100
Epoch[092] Batch [2399]/[3760]	Speed: 69.626705 samples/sec	 accuracy=70.758464	 loss=1.186125	 lr=0.000100
Epoch[092] Batch [2449]/[3760]	Speed: 69.681049 samples/sec	 accuracy=70.767219	 loss=1.185127	 lr=0.000100
Epoch[092] Batch [2499]/[3760]	Speed: 70.302226 samples/sec	 accuracy=70.765000	 loss=1.185418	 lr=0.000100
Epoch[092] Batch [2549]/[3760]	Speed: 68.492713 samples/sec	 accuracy=70.786152	 loss=1.184435	 lr=0.000100
Epoch[092] Batch [2599]/[3760]	Speed: 69.709375 samples/sec	 accuracy=70.780048	 loss=1.184529	 lr=0.000100
Epoch[092] Batch [2649]/[3760]	Speed: 68.697468 samples/sec	 accuracy=70.788325	 loss=1.184025	 lr=0.000100
Epoch[092] Batch [2699]/[3760]	Speed: 68.828501 samples/sec	 accuracy=70.776620	 loss=1.184341	 lr=0.000100
Epoch[092] Batch [2749]/[3760]	Speed: 70.129334 samples/sec	 accuracy=70.780682	 loss=1.184142	 lr=0.000100
Epoch[092] Batch [2799]/[3760]	Speed: 71.188031 samples/sec	 accuracy=70.773438	 loss=1.184257	 lr=0.000100
Epoch[092] Batch [2849]/[3760]	Speed: 69.317603 samples/sec	 accuracy=70.776864	 loss=1.183706	 lr=0.000100
Epoch[092] Batch [2899]/[3760]	Speed: 69.407141 samples/sec	 accuracy=70.778017	 loss=1.183723	 lr=0.000100
Epoch[092] Batch [2949]/[3760]	Speed: 70.546312 samples/sec	 accuracy=70.741525	 loss=1.185037	 lr=0.000100
Epoch[092] Batch [2999]/[3760]	Speed: 69.066760 samples/sec	 accuracy=70.747917	 loss=1.185235	 lr=0.000100
Epoch[092] Batch [3049]/[3760]	Speed: 70.247024 samples/sec	 accuracy=70.750512	 loss=1.185078	 lr=0.000100
Epoch[092] Batch [3099]/[3760]	Speed: 69.609546 samples/sec	 accuracy=70.743448	 loss=1.184904	 lr=0.000100
Epoch[092] Batch [3149]/[3760]	Speed: 70.843267 samples/sec	 accuracy=70.730159	 loss=1.185108	 lr=0.000100
Epoch[092] Batch [3199]/[3760]	Speed: 68.857949 samples/sec	 accuracy=70.740723	 loss=1.184841	 lr=0.000100
Epoch[092] Batch [3249]/[3760]	Speed: 70.130008 samples/sec	 accuracy=70.733654	 loss=1.184935	 lr=0.000100
Epoch[092] Batch [3299]/[3760]	Speed: 68.271206 samples/sec	 accuracy=70.740530	 loss=1.184673	 lr=0.000100
Epoch[092] Batch [3349]/[3760]	Speed: 71.026462 samples/sec	 accuracy=70.721549	 loss=1.185549	 lr=0.000100
Epoch[092] Batch [3399]/[3760]	Speed: 69.288661 samples/sec	 accuracy=70.736673	 loss=1.184838	 lr=0.000100
Epoch[092] Batch [3449]/[3760]	Speed: 69.720428 samples/sec	 accuracy=70.724185	 loss=1.185356	 lr=0.000100
Epoch[092] Batch [3499]/[3760]	Speed: 69.936520 samples/sec	 accuracy=70.722768	 loss=1.185688	 lr=0.000100
Epoch[092] Batch [3549]/[3760]	Speed: 70.145870 samples/sec	 accuracy=70.724472	 loss=1.185745	 lr=0.000100
Epoch[092] Batch [3599]/[3760]	Speed: 68.169420 samples/sec	 accuracy=70.741319	 loss=1.185703	 lr=0.000100
Epoch[092] Batch [3649]/[3760]	Speed: 69.921426 samples/sec	 accuracy=70.750856	 loss=1.185532	 lr=0.000100
Epoch[092] Batch [3699]/[3760]	Speed: 70.302558 samples/sec	 accuracy=70.747466	 loss=1.185570	 lr=0.000100
Epoch[092] Batch [3749]/[3760]	Speed: 77.381992 samples/sec	 accuracy=70.740417	 loss=1.185378	 lr=0.000100
Batch [0049]/[0303]: acc-top1=62.968750 acc-top5=83.906250
Batch [0099]/[0303]: acc-top1=62.921875 acc-top5=84.281250
Batch [0149]/[0303]: acc-top1=63.979167 acc-top5=84.697917
Batch [0199]/[0303]: acc-top1=64.390625 acc-top5=84.992188
Batch [0249]/[0303]: acc-top1=64.662500 acc-top5=85.212500
Batch [0299]/[0303]: acc-top1=64.463542 acc-top5=85.104167
[Epoch 092] training: accuracy=70.745512	 loss=1.185205
[Epoch 092] speed: 69 samples/sec	time cost: 3754.950369
[Epoch 092] validation: acc-top1=64.454414 acc-top5=85.117574 loss=1.553254
Epoch[093] Batch [0049]/[3759]	Speed: 48.507912 samples/sec	 accuracy=70.906250	 loss=1.173038	 lr=0.000100
Epoch[093] Batch [0099]/[3759]	Speed: 68.289946 samples/sec	 accuracy=70.562500	 loss=1.178826	 lr=0.000100
Epoch[093] Batch [0149]/[3759]	Speed: 68.449004 samples/sec	 accuracy=70.635417	 loss=1.172669	 lr=0.000100
Epoch[093] Batch [0199]/[3759]	Speed: 70.245762 samples/sec	 accuracy=70.804688	 loss=1.175648	 lr=0.000100
Epoch[093] Batch [0249]/[3759]	Speed: 71.075003 samples/sec	 accuracy=70.775000	 loss=1.178734	 lr=0.000100
Epoch[093] Batch [0299]/[3759]	Speed: 69.493930 samples/sec	 accuracy=71.098958	 loss=1.168532	 lr=0.000100
Epoch[093] Batch [0349]/[3759]	Speed: 69.669835 samples/sec	 accuracy=70.924107	 loss=1.174388	 lr=0.000100
Epoch[093] Batch [0399]/[3759]	Speed: 71.039636 samples/sec	 accuracy=70.851562	 loss=1.178513	 lr=0.000100
Epoch[093] Batch [0449]/[3759]	Speed: 69.385064 samples/sec	 accuracy=70.864583	 loss=1.179706	 lr=0.000100
Epoch[093] Batch [0499]/[3759]	Speed: 69.445358 samples/sec	 accuracy=70.862500	 loss=1.180412	 lr=0.000100
Epoch[093] Batch [0549]/[3759]	Speed: 69.546464 samples/sec	 accuracy=70.860795	 loss=1.180034	 lr=0.000100
Epoch[093] Batch [0599]/[3759]	Speed: 69.230401 samples/sec	 accuracy=70.822917	 loss=1.181836	 lr=0.000100
Epoch[093] Batch [0649]/[3759]	Speed: 70.270068 samples/sec	 accuracy=70.750000	 loss=1.185381	 lr=0.000100
Epoch[093] Batch [0699]/[3759]	Speed: 70.303414 samples/sec	 accuracy=70.745536	 loss=1.184978	 lr=0.000100
Epoch[093] Batch [0749]/[3759]	Speed: 70.683532 samples/sec	 accuracy=70.787500	 loss=1.182112	 lr=0.000100
Epoch[093] Batch [0799]/[3759]	Speed: 69.090639 samples/sec	 accuracy=70.820312	 loss=1.180806	 lr=0.000100
Epoch[093] Batch [0849]/[3759]	Speed: 70.218298 samples/sec	 accuracy=70.919118	 loss=1.176916	 lr=0.000100
Epoch[093] Batch [0899]/[3759]	Speed: 69.519522 samples/sec	 accuracy=70.842014	 loss=1.178681	 lr=0.000100
Epoch[093] Batch [0949]/[3759]	Speed: 68.289354 samples/sec	 accuracy=70.820724	 loss=1.180172	 lr=0.000100
Epoch[093] Batch [0999]/[3759]	Speed: 70.544541 samples/sec	 accuracy=70.820312	 loss=1.179082	 lr=0.000100
Epoch[093] Batch [1049]/[3759]	Speed: 70.557081 samples/sec	 accuracy=70.825893	 loss=1.179068	 lr=0.000100
Epoch[093] Batch [1099]/[3759]	Speed: 67.807451 samples/sec	 accuracy=70.821023	 loss=1.179910	 lr=0.000100
Epoch[093] Batch [1149]/[3759]	Speed: 72.334298 samples/sec	 accuracy=70.760870	 loss=1.183225	 lr=0.000100
Epoch[093] Batch [1199]/[3759]	Speed: 68.960294 samples/sec	 accuracy=70.756510	 loss=1.183459	 lr=0.000100
Epoch[093] Batch [1249]/[3759]	Speed: 68.966783 samples/sec	 accuracy=70.731250	 loss=1.184068	 lr=0.000100
Epoch[093] Batch [1299]/[3759]	Speed: 70.584321 samples/sec	 accuracy=70.757212	 loss=1.182547	 lr=0.000100
Epoch[093] Batch [1349]/[3759]	Speed: 71.032444 samples/sec	 accuracy=70.773148	 loss=1.181215	 lr=0.000100
Epoch[093] Batch [1399]/[3759]	Speed: 68.283225 samples/sec	 accuracy=70.780134	 loss=1.181189	 lr=0.000100
Epoch[093] Batch [1449]/[3759]	Speed: 69.818016 samples/sec	 accuracy=70.798491	 loss=1.181590	 lr=0.000100
Epoch[093] Batch [1499]/[3759]	Speed: 69.679877 samples/sec	 accuracy=70.829167	 loss=1.180621	 lr=0.000100
Epoch[093] Batch [1549]/[3759]	Speed: 69.833210 samples/sec	 accuracy=70.808468	 loss=1.181037	 lr=0.000100
Epoch[093] Batch [1599]/[3759]	Speed: 68.734784 samples/sec	 accuracy=70.852539	 loss=1.178690	 lr=0.000100
Epoch[093] Batch [1649]/[3759]	Speed: 68.924846 samples/sec	 accuracy=70.842803	 loss=1.178365	 lr=0.000100
Epoch[093] Batch [1699]/[3759]	Speed: 69.467326 samples/sec	 accuracy=70.853860	 loss=1.178365	 lr=0.000100
Epoch[093] Batch [1749]/[3759]	Speed: 70.671680 samples/sec	 accuracy=70.891964	 loss=1.176564	 lr=0.000100
Epoch[093] Batch [1799]/[3759]	Speed: 66.941931 samples/sec	 accuracy=70.881076	 loss=1.176801	 lr=0.000100
Epoch[093] Batch [1849]/[3759]	Speed: 71.577408 samples/sec	 accuracy=70.864020	 loss=1.177990	 lr=0.000100
Epoch[093] Batch [1899]/[3759]	Speed: 70.375419 samples/sec	 accuracy=70.876645	 loss=1.177981	 lr=0.000100
Epoch[093] Batch [1949]/[3759]	Speed: 68.627259 samples/sec	 accuracy=70.852564	 loss=1.178787	 lr=0.000100
Epoch[093] Batch [1999]/[3759]	Speed: 71.288597 samples/sec	 accuracy=70.850000	 loss=1.179033	 lr=0.000100
Epoch[093] Batch [2049]/[3759]	Speed: 69.195013 samples/sec	 accuracy=70.842988	 loss=1.179435	 lr=0.000100
Epoch[093] Batch [2099]/[3759]	Speed: 70.509771 samples/sec	 accuracy=70.831845	 loss=1.179849	 lr=0.000100
Epoch[093] Batch [2149]/[3759]	Speed: 70.709354 samples/sec	 accuracy=70.799419	 loss=1.180931	 lr=0.000100
Epoch[093] Batch [2199]/[3759]	Speed: 68.766779 samples/sec	 accuracy=70.800426	 loss=1.180617	 lr=0.000100
Epoch[093] Batch [2249]/[3759]	Speed: 70.714539 samples/sec	 accuracy=70.822917	 loss=1.179530	 lr=0.000100
Epoch[093] Batch [2299]/[3759]	Speed: 67.817007 samples/sec	 accuracy=70.794158	 loss=1.181348	 lr=0.000100
Epoch[093] Batch [2349]/[3759]	Speed: 70.426376 samples/sec	 accuracy=70.805186	 loss=1.181551	 lr=0.000100
Epoch[093] Batch [2399]/[3759]	Speed: 69.696075 samples/sec	 accuracy=70.791667	 loss=1.182077	 lr=0.000100
Epoch[093] Batch [2449]/[3759]	Speed: 70.780392 samples/sec	 accuracy=70.759566	 loss=1.182915	 lr=0.000100
Epoch[093] Batch [2499]/[3759]	Speed: 69.281280 samples/sec	 accuracy=70.753750	 loss=1.183413	 lr=0.000100
Epoch[093] Batch [2549]/[3759]	Speed: 69.984353 samples/sec	 accuracy=70.750000	 loss=1.183308	 lr=0.000100
Epoch[093] Batch [2599]/[3759]	Speed: 70.095377 samples/sec	 accuracy=70.778245	 loss=1.182127	 lr=0.000100
Epoch[093] Batch [2649]/[3759]	Speed: 68.572882 samples/sec	 accuracy=70.784788	 loss=1.181822	 lr=0.000100
Epoch[093] Batch [2699]/[3759]	Speed: 68.624161 samples/sec	 accuracy=70.774884	 loss=1.182385	 lr=0.000100
Epoch[093] Batch [2749]/[3759]	Speed: 70.940892 samples/sec	 accuracy=70.769318	 loss=1.182480	 lr=0.000100
Epoch[093] Batch [2799]/[3759]	Speed: 70.072092 samples/sec	 accuracy=70.783482	 loss=1.181945	 lr=0.000100
Epoch[093] Batch [2849]/[3759]	Speed: 69.916038 samples/sec	 accuracy=70.763706	 loss=1.182125	 lr=0.000100
Epoch[093] Batch [2899]/[3759]	Speed: 71.026570 samples/sec	 accuracy=70.779095	 loss=1.181640	 lr=0.000100
Epoch[093] Batch [2949]/[3759]	Speed: 69.248127 samples/sec	 accuracy=70.767479	 loss=1.181652	 lr=0.000100
Epoch[093] Batch [2999]/[3759]	Speed: 68.667496 samples/sec	 accuracy=70.732292	 loss=1.182831	 lr=0.000100
Epoch[093] Batch [3049]/[3759]	Speed: 70.894210 samples/sec	 accuracy=70.735656	 loss=1.182871	 lr=0.000100
Epoch[093] Batch [3099]/[3759]	Speed: 69.203388 samples/sec	 accuracy=70.742944	 loss=1.182870	 lr=0.000100
Epoch[093] Batch [3149]/[3759]	Speed: 69.069527 samples/sec	 accuracy=70.750000	 loss=1.182477	 lr=0.000100
Epoch[093] Batch [3199]/[3759]	Speed: 69.826534 samples/sec	 accuracy=70.739746	 loss=1.183193	 lr=0.000100
Epoch[093] Batch [3249]/[3759]	Speed: 68.715104 samples/sec	 accuracy=70.740385	 loss=1.182509	 lr=0.000100
Epoch[093] Batch [3299]/[3759]	Speed: 69.552220 samples/sec	 accuracy=70.757576	 loss=1.181854	 lr=0.000100
Epoch[093] Batch [3349]/[3759]	Speed: 70.543910 samples/sec	 accuracy=70.747201	 loss=1.182224	 lr=0.000100
Epoch[093] Batch [3399]/[3759]	Speed: 70.796810 samples/sec	 accuracy=70.751838	 loss=1.182066	 lr=0.000100
Epoch[093] Batch [3449]/[3759]	Speed: 68.904964 samples/sec	 accuracy=70.751359	 loss=1.182143	 lr=0.000100
Epoch[093] Batch [3499]/[3759]	Speed: 69.032244 samples/sec	 accuracy=70.749107	 loss=1.182124	 lr=0.000100
Epoch[093] Batch [3549]/[3759]	Speed: 70.556313 samples/sec	 accuracy=70.744278	 loss=1.181964	 lr=0.000100
Epoch[093] Batch [3599]/[3759]	Speed: 70.778207 samples/sec	 accuracy=70.740017	 loss=1.182180	 lr=0.000100
Epoch[093] Batch [3649]/[3759]	Speed: 68.562323 samples/sec	 accuracy=70.723031	 loss=1.183176	 lr=0.000100
Epoch[093] Batch [3699]/[3759]	Speed: 70.642123 samples/sec	 accuracy=70.730574	 loss=1.182877	 lr=0.000100
Epoch[093] Batch [3749]/[3759]	Speed: 76.663744 samples/sec	 accuracy=70.720000	 loss=1.183411	 lr=0.000100
Batch [0049]/[0303]: acc-top1=62.875000 acc-top5=84.312500
Batch [0099]/[0303]: acc-top1=63.000000 acc-top5=84.359375
Batch [0149]/[0303]: acc-top1=63.822917 acc-top5=84.812500
Batch [0199]/[0303]: acc-top1=64.375000 acc-top5=85.164062
Batch [0249]/[0303]: acc-top1=64.687500 acc-top5=85.375000
Batch [0299]/[0303]: acc-top1=64.447917 acc-top5=85.244792
[Epoch 093] training: accuracy=70.716530	 loss=1.183289
[Epoch 093] speed: 69 samples/sec	time cost: 3748.884131
[Epoch 093] validation: acc-top1=64.413160 acc-top5=85.256807 loss=1.551487
Epoch[094] Batch [0049]/[3760]	Speed: 47.520696 samples/sec	 accuracy=69.531250	 loss=1.235594	 lr=0.000100
Epoch[094] Batch [0099]/[3760]	Speed: 69.429382 samples/sec	 accuracy=70.187500	 loss=1.204253	 lr=0.000100
Epoch[094] Batch [0149]/[3760]	Speed: 68.278308 samples/sec	 accuracy=70.281250	 loss=1.193896	 lr=0.000100
Epoch[094] Batch [0199]/[3760]	Speed: 70.323599 samples/sec	 accuracy=70.156250	 loss=1.198126	 lr=0.000100
Epoch[094] Batch [0249]/[3760]	Speed: 69.067453 samples/sec	 accuracy=70.437500	 loss=1.190714	 lr=0.000100
Epoch[094] Batch [0299]/[3760]	Speed: 69.338849 samples/sec	 accuracy=70.265625	 loss=1.196999	 lr=0.000100
Epoch[094] Batch [0349]/[3760]	Speed: 68.858511 samples/sec	 accuracy=70.308036	 loss=1.196629	 lr=0.000100
Epoch[094] Batch [0399]/[3760]	Speed: 70.778775 samples/sec	 accuracy=70.414062	 loss=1.189946	 lr=0.000100
Epoch[094] Batch [0449]/[3760]	Speed: 69.082880 samples/sec	 accuracy=70.423611	 loss=1.187367	 lr=0.000100
Epoch[094] Batch [0499]/[3760]	Speed: 70.665515 samples/sec	 accuracy=70.409375	 loss=1.188610	 lr=0.000100
Epoch[094] Batch [0549]/[3760]	Speed: 68.863385 samples/sec	 accuracy=70.369318	 loss=1.190216	 lr=0.000100
Epoch[094] Batch [0599]/[3760]	Speed: 70.258709 samples/sec	 accuracy=70.380208	 loss=1.189203	 lr=0.000100
Epoch[094] Batch [0649]/[3760]	Speed: 69.060800 samples/sec	 accuracy=70.382212	 loss=1.189583	 lr=0.000100
Epoch[094] Batch [0699]/[3760]	Speed: 69.229900 samples/sec	 accuracy=70.352679	 loss=1.189835	 lr=0.000100
Epoch[094] Batch [0749]/[3760]	Speed: 70.495335 samples/sec	 accuracy=70.447917	 loss=1.185205	 lr=0.000100
Epoch[094] Batch [0799]/[3760]	Speed: 67.673890 samples/sec	 accuracy=70.523438	 loss=1.182395	 lr=0.000100
Epoch[094] Batch [0849]/[3760]	Speed: 67.918374 samples/sec	 accuracy=70.534926	 loss=1.182507	 lr=0.000100
Epoch[094] Batch [0899]/[3760]	Speed: 72.349166 samples/sec	 accuracy=70.486111	 loss=1.183519	 lr=0.000100
Epoch[094] Batch [0949]/[3760]	Speed: 69.183516 samples/sec	 accuracy=70.550987	 loss=1.182815	 lr=0.000100
Epoch[094] Batch [0999]/[3760]	Speed: 70.081972 samples/sec	 accuracy=70.556250	 loss=1.181751	 lr=0.000100
Epoch[094] Batch [1049]/[3760]	Speed: 69.881639 samples/sec	 accuracy=70.556548	 loss=1.182119	 lr=0.000100
Epoch[094] Batch [1099]/[3760]	Speed: 65.933112 samples/sec	 accuracy=70.525568	 loss=1.183101	 lr=0.000100
Epoch[094] Batch [1149]/[3760]	Speed: 72.582839 samples/sec	 accuracy=70.532609	 loss=1.183475	 lr=0.000100
Epoch[094] Batch [1199]/[3760]	Speed: 70.135978 samples/sec	 accuracy=70.492188	 loss=1.184533	 lr=0.000100
Epoch[094] Batch [1249]/[3760]	Speed: 68.473306 samples/sec	 accuracy=70.493750	 loss=1.184717	 lr=0.000100
Epoch[094] Batch [1299]/[3760]	Speed: 71.834298 samples/sec	 accuracy=70.514423	 loss=1.185433	 lr=0.000100
Epoch[094] Batch [1349]/[3760]	Speed: 70.685649 samples/sec	 accuracy=70.557870	 loss=1.184936	 lr=0.000100
Epoch[094] Batch [1399]/[3760]	Speed: 66.234207 samples/sec	 accuracy=70.523438	 loss=1.186037	 lr=0.000100
Epoch[094] Batch [1449]/[3760]	Speed: 71.714291 samples/sec	 accuracy=70.538793	 loss=1.185429	 lr=0.000100
Epoch[094] Batch [1499]/[3760]	Speed: 65.612396 samples/sec	 accuracy=70.557292	 loss=1.184950	 lr=0.000100
Epoch[094] Batch [1549]/[3760]	Speed: 73.604337 samples/sec	 accuracy=70.530242	 loss=1.185647	 lr=0.000100
Epoch[094] Batch [1599]/[3760]	Speed: 70.118386 samples/sec	 accuracy=70.531250	 loss=1.185322	 lr=0.000100
Epoch[094] Batch [1649]/[3760]	Speed: 70.039955 samples/sec	 accuracy=70.550189	 loss=1.185418	 lr=0.000100
Epoch[094] Batch [1699]/[3760]	Speed: 68.991830 samples/sec	 accuracy=70.537684	 loss=1.185797	 lr=0.000100
Epoch[094] Batch [1749]/[3760]	Speed: 69.186644 samples/sec	 accuracy=70.525893	 loss=1.186154	 lr=0.000100
Epoch[094] Batch [1799]/[3760]	Speed: 66.904519 samples/sec	 accuracy=70.496528	 loss=1.186667	 lr=0.000100
Epoch[094] Batch [1849]/[3760]	Speed: 72.644856 samples/sec	 accuracy=70.480574	 loss=1.187152	 lr=0.000100
Epoch[094] Batch [1899]/[3760]	Speed: 70.296896 samples/sec	 accuracy=70.509046	 loss=1.185960	 lr=0.000100
Epoch[094] Batch [1949]/[3760]	Speed: 67.894778 samples/sec	 accuracy=70.559295	 loss=1.184149	 lr=0.000100
Epoch[094] Batch [1999]/[3760]	Speed: 70.443950 samples/sec	 accuracy=70.576562	 loss=1.184459	 lr=0.000100
Epoch[094] Batch [2049]/[3760]	Speed: 68.390450 samples/sec	 accuracy=70.570884	 loss=1.184582	 lr=0.000100
Epoch[094] Batch [2099]/[3760]	Speed: 70.887182 samples/sec	 accuracy=70.572173	 loss=1.185409	 lr=0.000100
Epoch[094] Batch [2149]/[3760]	Speed: 68.578557 samples/sec	 accuracy=70.601744	 loss=1.184211	 lr=0.000100
Epoch[094] Batch [2199]/[3760]	Speed: 70.534678 samples/sec	 accuracy=70.580966	 loss=1.184749	 lr=0.000100
Epoch[094] Batch [2249]/[3760]	Speed: 69.884721 samples/sec	 accuracy=70.583333	 loss=1.184880	 lr=0.000100
Epoch[094] Batch [2299]/[3760]	Speed: 70.870416 samples/sec	 accuracy=70.595109	 loss=1.184001	 lr=0.000100
Epoch[094] Batch [2349]/[3760]	Speed: 69.422778 samples/sec	 accuracy=70.582447	 loss=1.183719	 lr=0.000100
Epoch[094] Batch [2399]/[3760]	Speed: 69.383570 samples/sec	 accuracy=70.569010	 loss=1.183824	 lr=0.000100
Epoch[094] Batch [2449]/[3760]	Speed: 69.768362 samples/sec	 accuracy=70.579719	 loss=1.183681	 lr=0.000100
Epoch[094] Batch [2499]/[3760]	Speed: 68.570732 samples/sec	 accuracy=70.578125	 loss=1.183783	 lr=0.000100
Epoch[094] Batch [2549]/[3760]	Speed: 70.445396 samples/sec	 accuracy=70.591912	 loss=1.183544	 lr=0.000100
Epoch[094] Batch [2599]/[3760]	Speed: 69.802858 samples/sec	 accuracy=70.592548	 loss=1.182980	 lr=0.000100
Epoch[094] Batch [2649]/[3760]	Speed: 68.487678 samples/sec	 accuracy=70.603184	 loss=1.182832	 lr=0.000100
Epoch[094] Batch [2699]/[3760]	Speed: 69.878074 samples/sec	 accuracy=70.619213	 loss=1.181740	 lr=0.000100
Epoch[094] Batch [2749]/[3760]	Speed: 71.528707 samples/sec	 accuracy=70.642045	 loss=1.180914	 lr=0.000100
Epoch[094] Batch [2799]/[3760]	Speed: 69.614003 samples/sec	 accuracy=70.636161	 loss=1.181134	 lr=0.000100
Epoch[094] Batch [2849]/[3760]	Speed: 69.868510 samples/sec	 accuracy=70.646382	 loss=1.180491	 lr=0.000100
Epoch[094] Batch [2899]/[3760]	Speed: 68.293710 samples/sec	 accuracy=70.647629	 loss=1.179605	 lr=0.000100
Epoch[094] Batch [2949]/[3760]	Speed: 70.336534 samples/sec	 accuracy=70.642479	 loss=1.180576	 lr=0.000100
Epoch[094] Batch [2999]/[3760]	Speed: 70.119395 samples/sec	 accuracy=70.662500	 loss=1.179891	 lr=0.000100
Epoch[094] Batch [3049]/[3760]	Speed: 65.412396 samples/sec	 accuracy=70.667008	 loss=1.180047	 lr=0.000100
Epoch[094] Batch [3099]/[3760]	Speed: 72.268239 samples/sec	 accuracy=70.668851	 loss=1.180138	 lr=0.000100
Epoch[094] Batch [3149]/[3760]	Speed: 69.664247 samples/sec	 accuracy=70.679563	 loss=1.180292	 lr=0.000100
Epoch[094] Batch [3199]/[3760]	Speed: 70.391664 samples/sec	 accuracy=70.659668	 loss=1.181047	 lr=0.000100
Epoch[094] Batch [3249]/[3760]	Speed: 70.189722 samples/sec	 accuracy=70.675000	 loss=1.180382	 lr=0.000100
Epoch[094] Batch [3299]/[3760]	Speed: 67.945038 samples/sec	 accuracy=70.666193	 loss=1.180729	 lr=0.000100
Epoch[094] Batch [3349]/[3760]	Speed: 70.391069 samples/sec	 accuracy=70.662313	 loss=1.181010	 lr=0.000100
Epoch[094] Batch [3399]/[3760]	Speed: 69.849198 samples/sec	 accuracy=70.651654	 loss=1.181836	 lr=0.000100
Epoch[094] Batch [3449]/[3760]	Speed: 70.599424 samples/sec	 accuracy=70.665761	 loss=1.181273	 lr=0.000100
Epoch[094] Batch [3499]/[3760]	Speed: 69.037504 samples/sec	 accuracy=70.672768	 loss=1.181454	 lr=0.000100
Epoch[094] Batch [3549]/[3760]	Speed: 69.480228 samples/sec	 accuracy=70.672535	 loss=1.181654	 lr=0.000100
Epoch[094] Batch [3599]/[3760]	Speed: 69.538172 samples/sec	 accuracy=70.658420	 loss=1.182382	 lr=0.000100
Epoch[094] Batch [3649]/[3760]	Speed: 70.223806 samples/sec	 accuracy=70.666096	 loss=1.181942	 lr=0.000100
Epoch[094] Batch [3699]/[3760]	Speed: 70.330887 samples/sec	 accuracy=70.654561	 loss=1.182205	 lr=0.000100
Epoch[094] Batch [3749]/[3760]	Speed: 76.529655 samples/sec	 accuracy=70.651250	 loss=1.182518	 lr=0.000100
Batch [0049]/[0303]: acc-top1=63.125000 acc-top5=84.125000
Batch [0099]/[0303]: acc-top1=63.125000 acc-top5=84.140625
Batch [0149]/[0303]: acc-top1=64.031250 acc-top5=84.489583
Batch [0199]/[0303]: acc-top1=64.531250 acc-top5=84.882812
Batch [0249]/[0303]: acc-top1=64.775000 acc-top5=85.156250
Batch [0299]/[0303]: acc-top1=64.562500 acc-top5=85.010417
[Epoch 094] training: accuracy=70.648687	 loss=1.182540
[Epoch 094] speed: 69 samples/sec	time cost: 3756.542732
[Epoch 094] validation: acc-top1=64.536922 acc-top5=85.035066 loss=1.552938
Epoch[095] Batch [0049]/[3760]	Speed: 48.988006 samples/sec	 accuracy=71.343750	 loss=1.167556	 lr=0.000100
Epoch[095] Batch [0099]/[3760]	Speed: 67.652907 samples/sec	 accuracy=71.140625	 loss=1.177008	 lr=0.000100
Epoch[095] Batch [0149]/[3760]	Speed: 69.191464 samples/sec	 accuracy=70.552083	 loss=1.187252	 lr=0.000100
Epoch[095] Batch [0199]/[3760]	Speed: 68.269056 samples/sec	 accuracy=70.468750	 loss=1.200061	 lr=0.000100
Epoch[095] Batch [0249]/[3760]	Speed: 71.068260 samples/sec	 accuracy=70.550000	 loss=1.196774	 lr=0.000100
Epoch[095] Batch [0299]/[3760]	Speed: 69.606968 samples/sec	 accuracy=70.432292	 loss=1.199967	 lr=0.000100
Epoch[095] Batch [0349]/[3760]	Speed: 70.387466 samples/sec	 accuracy=70.602679	 loss=1.192284	 lr=0.000100
Epoch[095] Batch [0399]/[3760]	Speed: 67.460416 samples/sec	 accuracy=70.632812	 loss=1.190305	 lr=0.000100
Epoch[095] Batch [0449]/[3760]	Speed: 70.756084 samples/sec	 accuracy=70.753472	 loss=1.182046	 lr=0.000100
Epoch[095] Batch [0499]/[3760]	Speed: 70.722697 samples/sec	 accuracy=70.715625	 loss=1.182306	 lr=0.000100
Epoch[095] Batch [0549]/[3760]	Speed: 68.587374 samples/sec	 accuracy=70.653409	 loss=1.182808	 lr=0.000100
Epoch[095] Batch [0599]/[3760]	Speed: 71.237953 samples/sec	 accuracy=70.614583	 loss=1.184199	 lr=0.000100
Epoch[095] Batch [0649]/[3760]	Speed: 68.263630 samples/sec	 accuracy=70.584135	 loss=1.186579	 lr=0.000100
Epoch[095] Batch [0699]/[3760]	Speed: 67.968061 samples/sec	 accuracy=70.647321	 loss=1.184819	 lr=0.000100
Epoch[095] Batch [0749]/[3760]	Speed: 69.324402 samples/sec	 accuracy=70.647917	 loss=1.185922	 lr=0.000100
Epoch[095] Batch [0799]/[3760]	Speed: 70.638513 samples/sec	 accuracy=70.667969	 loss=1.184151	 lr=0.000100
Epoch[095] Batch [0849]/[3760]	Speed: 66.045453 samples/sec	 accuracy=70.724265	 loss=1.181866	 lr=0.000100
Epoch[095] Batch [0899]/[3760]	Speed: 72.987333 samples/sec	 accuracy=70.699653	 loss=1.181964	 lr=0.000100
Epoch[095] Batch [0949]/[3760]	Speed: 70.160706 samples/sec	 accuracy=70.664474	 loss=1.182828	 lr=0.000100
Epoch[095] Batch [0999]/[3760]	Speed: 69.593808 samples/sec	 accuracy=70.690625	 loss=1.183165	 lr=0.000100
Epoch[095] Batch [1049]/[3760]	Speed: 69.109435 samples/sec	 accuracy=70.711310	 loss=1.182499	 lr=0.000100
Epoch[095] Batch [1099]/[3760]	Speed: 69.316436 samples/sec	 accuracy=70.710227	 loss=1.182524	 lr=0.000100
Epoch[095] Batch [1149]/[3760]	Speed: 69.233789 samples/sec	 accuracy=70.623641	 loss=1.186099	 lr=0.000100
Epoch[095] Batch [1199]/[3760]	Speed: 69.168310 samples/sec	 accuracy=70.579427	 loss=1.187542	 lr=0.000100
Epoch[095] Batch [1249]/[3760]	Speed: 71.215155 samples/sec	 accuracy=70.596250	 loss=1.188505	 lr=0.000100
Epoch[095] Batch [1299]/[3760]	Speed: 69.169568 samples/sec	 accuracy=70.575721	 loss=1.190134	 lr=0.000100
Epoch[095] Batch [1349]/[3760]	Speed: 68.829570 samples/sec	 accuracy=70.584491	 loss=1.189545	 lr=0.000100
Epoch[095] Batch [1399]/[3760]	Speed: 68.985126 samples/sec	 accuracy=70.579241	 loss=1.189205	 lr=0.000100
Epoch[095] Batch [1449]/[3760]	Speed: 70.044813 samples/sec	 accuracy=70.549569	 loss=1.190182	 lr=0.000100
Epoch[095] Batch [1499]/[3760]	Speed: 70.526920 samples/sec	 accuracy=70.567708	 loss=1.189449	 lr=0.000100
Epoch[095] Batch [1549]/[3760]	Speed: 69.165066 samples/sec	 accuracy=70.588710	 loss=1.188915	 lr=0.000100
Epoch[095] Batch [1599]/[3760]	Speed: 69.703854 samples/sec	 accuracy=70.614258	 loss=1.188563	 lr=0.000100
Epoch[095] Batch [1649]/[3760]	Speed: 69.951098 samples/sec	 accuracy=70.615530	 loss=1.189574	 lr=0.000100
Epoch[095] Batch [1699]/[3760]	Speed: 68.110630 samples/sec	 accuracy=70.630515	 loss=1.189141	 lr=0.000100
Epoch[095] Batch [1749]/[3760]	Speed: 69.141787 samples/sec	 accuracy=70.650893	 loss=1.187883	 lr=0.000100
Epoch[095] Batch [1799]/[3760]	Speed: 68.741542 samples/sec	 accuracy=70.633681	 loss=1.188274	 lr=0.000100
Epoch[095] Batch [1849]/[3760]	Speed: 71.577410 samples/sec	 accuracy=70.636824	 loss=1.188849	 lr=0.000100
Epoch[095] Batch [1899]/[3760]	Speed: 69.941773 samples/sec	 accuracy=70.642270	 loss=1.188451	 lr=0.000100
Epoch[095] Batch [1949]/[3760]	Speed: 67.416145 samples/sec	 accuracy=70.626603	 loss=1.189163	 lr=0.000100
Epoch[095] Batch [1999]/[3760]	Speed: 70.402135 samples/sec	 accuracy=70.590625	 loss=1.189558	 lr=0.000100
Epoch[095] Batch [2049]/[3760]	Speed: 70.679108 samples/sec	 accuracy=70.589177	 loss=1.189306	 lr=0.000100
Epoch[095] Batch [2099]/[3760]	Speed: 71.015573 samples/sec	 accuracy=70.601190	 loss=1.189191	 lr=0.000100
Epoch[095] Batch [2149]/[3760]	Speed: 66.190783 samples/sec	 accuracy=70.609012	 loss=1.189015	 lr=0.000100
Epoch[095] Batch [2199]/[3760]	Speed: 72.969805 samples/sec	 accuracy=70.584517	 loss=1.189853	 lr=0.000100
Epoch[095] Batch [2249]/[3760]	Speed: 69.140448 samples/sec	 accuracy=70.595833	 loss=1.189176	 lr=0.000100
Epoch[095] Batch [2299]/[3760]	Speed: 69.307277 samples/sec	 accuracy=70.582880	 loss=1.189571	 lr=0.000100
Epoch[095] Batch [2349]/[3760]	Speed: 69.245603 samples/sec	 accuracy=70.579122	 loss=1.189092	 lr=0.000100
Epoch[095] Batch [2399]/[3760]	Speed: 69.943609 samples/sec	 accuracy=70.596354	 loss=1.187989	 lr=0.000100
Epoch[095] Batch [2449]/[3760]	Speed: 70.317200 samples/sec	 accuracy=70.588010	 loss=1.187747	 lr=0.000100
Epoch[095] Batch [2499]/[3760]	Speed: 69.015775 samples/sec	 accuracy=70.569375	 loss=1.188607	 lr=0.000100
Epoch[095] Batch [2549]/[3760]	Speed: 69.751053 samples/sec	 accuracy=70.570466	 loss=1.187987	 lr=0.000100
Epoch[095] Batch [2599]/[3760]	Speed: 68.657746 samples/sec	 accuracy=70.595553	 loss=1.187514	 lr=0.000100
Epoch[095] Batch [2649]/[3760]	Speed: 68.984925 samples/sec	 accuracy=70.616745	 loss=1.186444	 lr=0.000100
Epoch[095] Batch [2699]/[3760]	Speed: 71.292895 samples/sec	 accuracy=70.593750	 loss=1.187210	 lr=0.000100
Epoch[095] Batch [2749]/[3760]	Speed: 68.225379 samples/sec	 accuracy=70.598295	 loss=1.187123	 lr=0.000100
Epoch[095] Batch [2799]/[3760]	Speed: 70.728381 samples/sec	 accuracy=70.618862	 loss=1.186297	 lr=0.000100
Epoch[095] Batch [2849]/[3760]	Speed: 70.617295 samples/sec	 accuracy=70.598136	 loss=1.186477	 lr=0.000100
Epoch[095] Batch [2899]/[3760]	Speed: 70.186538 samples/sec	 accuracy=70.585668	 loss=1.186783	 lr=0.000100
Epoch[095] Batch [2949]/[3760]	Speed: 68.082714 samples/sec	 accuracy=70.582097	 loss=1.187106	 lr=0.000100
Epoch[095] Batch [2999]/[3760]	Speed: 68.591128 samples/sec	 accuracy=70.586979	 loss=1.186988	 lr=0.000100
Epoch[095] Batch [3049]/[3760]	Speed: 71.370147 samples/sec	 accuracy=70.609119	 loss=1.186604	 lr=0.000100
Epoch[095] Batch [3099]/[3760]	Speed: 70.313618 samples/sec	 accuracy=70.621976	 loss=1.186316	 lr=0.000100
Epoch[095] Batch [3149]/[3760]	Speed: 69.223308 samples/sec	 accuracy=70.613095	 loss=1.186436	 lr=0.000100
Epoch[095] Batch [3199]/[3760]	Speed: 69.726618 samples/sec	 accuracy=70.614746	 loss=1.186748	 lr=0.000100
Epoch[095] Batch [3249]/[3760]	Speed: 70.403388 samples/sec	 accuracy=70.607692	 loss=1.186726	 lr=0.000100
Epoch[095] Batch [3299]/[3760]	Speed: 68.621789 samples/sec	 accuracy=70.625473	 loss=1.186355	 lr=0.000100
Epoch[095] Batch [3349]/[3760]	Speed: 70.207909 samples/sec	 accuracy=70.614739	 loss=1.186471	 lr=0.000100
Epoch[095] Batch [3399]/[3760]	Speed: 69.252231 samples/sec	 accuracy=70.608456	 loss=1.186001	 lr=0.000100
Epoch[095] Batch [3449]/[3760]	Speed: 68.778591 samples/sec	 accuracy=70.613678	 loss=1.185660	 lr=0.000100
Epoch[095] Batch [3499]/[3760]	Speed: 71.352881 samples/sec	 accuracy=70.590625	 loss=1.186050	 lr=0.000100
Epoch[095] Batch [3549]/[3760]	Speed: 70.472631 samples/sec	 accuracy=70.611796	 loss=1.185599	 lr=0.000100
Epoch[095] Batch [3599]/[3760]	Speed: 71.046294 samples/sec	 accuracy=70.606337	 loss=1.185333	 lr=0.000100
Epoch[095] Batch [3649]/[3760]	Speed: 68.006790 samples/sec	 accuracy=70.608733	 loss=1.185172	 lr=0.000100
Epoch[095] Batch [3699]/[3760]	Speed: 70.663539 samples/sec	 accuracy=70.596706	 loss=1.185191	 lr=0.000100
Epoch[095] Batch [3749]/[3760]	Speed: 76.956306 samples/sec	 accuracy=70.595833	 loss=1.185742	 lr=0.000100
Batch [0049]/[0303]: acc-top1=63.093750 acc-top5=84.062500
Batch [0099]/[0303]: acc-top1=63.156250 acc-top5=84.171875
Batch [0149]/[0303]: acc-top1=64.062500 acc-top5=84.437500
Batch [0199]/[0303]: acc-top1=64.648438 acc-top5=84.812500
Batch [0249]/[0303]: acc-top1=64.793750 acc-top5=85.081250
Batch [0299]/[0303]: acc-top1=64.609375 acc-top5=84.994792
[Epoch 095] training: accuracy=70.604222	 loss=1.185582
[Epoch 095] speed: 69 samples/sec	time cost: 3756.270957
[Epoch 095] validation: acc-top1=64.557550 acc-top5=85.014439 loss=1.552830
Epoch[096] Batch [0049]/[3759]	Speed: 48.144405 samples/sec	 accuracy=71.281250	 loss=1.156657	 lr=0.000100
Epoch[096] Batch [0099]/[3759]	Speed: 68.825863 samples/sec	 accuracy=71.421875	 loss=1.149091	 lr=0.000100
Epoch[096] Batch [0149]/[3759]	Speed: 70.262905 samples/sec	 accuracy=71.406250	 loss=1.144474	 lr=0.000100
Epoch[096] Batch [0199]/[3759]	Speed: 69.823430 samples/sec	 accuracy=71.304688	 loss=1.152259	 lr=0.000100
Epoch[096] Batch [0249]/[3759]	Speed: 69.568303 samples/sec	 accuracy=71.018750	 loss=1.165920	 lr=0.000100
Epoch[096] Batch [0299]/[3759]	Speed: 70.820142 samples/sec	 accuracy=70.911458	 loss=1.169480	 lr=0.000100
Epoch[096] Batch [0349]/[3759]	Speed: 68.261834 samples/sec	 accuracy=71.017857	 loss=1.170370	 lr=0.000100
Epoch[096] Batch [0399]/[3759]	Speed: 70.568672 samples/sec	 accuracy=70.957031	 loss=1.174109	 lr=0.000100
Epoch[096] Batch [0449]/[3759]	Speed: 69.377897 samples/sec	 accuracy=70.885417	 loss=1.177983	 lr=0.000100
Epoch[096] Batch [0499]/[3759]	Speed: 70.948987 samples/sec	 accuracy=70.681250	 loss=1.181651	 lr=0.000100
Epoch[096] Batch [0549]/[3759]	Speed: 68.887827 samples/sec	 accuracy=70.633523	 loss=1.180622	 lr=0.000100
Epoch[096] Batch [0599]/[3759]	Speed: 70.361403 samples/sec	 accuracy=70.643229	 loss=1.179196	 lr=0.000100
Epoch[096] Batch [0649]/[3759]	Speed: 70.613112 samples/sec	 accuracy=70.740385	 loss=1.176190	 lr=0.000100
Epoch[096] Batch [0699]/[3759]	Speed: 68.879428 samples/sec	 accuracy=70.805804	 loss=1.173892	 lr=0.000100
Epoch[096] Batch [0749]/[3759]	Speed: 69.278439 samples/sec	 accuracy=70.793750	 loss=1.175592	 lr=0.000100
Epoch[096] Batch [0799]/[3759]	Speed: 69.971036 samples/sec	 accuracy=70.808594	 loss=1.176425	 lr=0.000100
Epoch[096] Batch [0849]/[3759]	Speed: 70.032354 samples/sec	 accuracy=70.816176	 loss=1.176351	 lr=0.000100
Epoch[096] Batch [0899]/[3759]	Speed: 66.851929 samples/sec	 accuracy=70.758681	 loss=1.179240	 lr=0.000100
Epoch[096] Batch [0949]/[3759]	Speed: 71.209443 samples/sec	 accuracy=70.773026	 loss=1.179694	 lr=0.000100
Epoch[096] Batch [0999]/[3759]	Speed: 70.198732 samples/sec	 accuracy=70.779687	 loss=1.178286	 lr=0.000100
Epoch[096] Batch [1049]/[3759]	Speed: 70.404989 samples/sec	 accuracy=70.776786	 loss=1.177480	 lr=0.000100
Epoch[096] Batch [1099]/[3759]	Speed: 68.974429 samples/sec	 accuracy=70.803977	 loss=1.177983	 lr=0.000100
Epoch[096] Batch [1149]/[3759]	Speed: 70.170471 samples/sec	 accuracy=70.789402	 loss=1.178454	 lr=0.000100
Epoch[096] Batch [1199]/[3759]	Speed: 69.873901 samples/sec	 accuracy=70.752604	 loss=1.179051	 lr=0.000100
Epoch[096] Batch [1249]/[3759]	Speed: 70.423504 samples/sec	 accuracy=70.757500	 loss=1.179823	 lr=0.000100
Epoch[096] Batch [1299]/[3759]	Speed: 69.989669 samples/sec	 accuracy=70.719952	 loss=1.181499	 lr=0.000100
Epoch[096] Batch [1349]/[3759]	Speed: 68.998057 samples/sec	 accuracy=70.714120	 loss=1.180382	 lr=0.000100
Epoch[096] Batch [1399]/[3759]	Speed: 67.907525 samples/sec	 accuracy=70.739955	 loss=1.180387	 lr=0.000100
Epoch[096] Batch [1449]/[3759]	Speed: 70.350897 samples/sec	 accuracy=70.785560	 loss=1.178879	 lr=0.000100
Epoch[096] Batch [1499]/[3759]	Speed: 67.862077 samples/sec	 accuracy=70.754167	 loss=1.179081	 lr=0.000100
Epoch[096] Batch [1549]/[3759]	Speed: 70.131592 samples/sec	 accuracy=70.767137	 loss=1.178415	 lr=0.000100
Epoch[096] Batch [1599]/[3759]	Speed: 71.560836 samples/sec	 accuracy=70.780273	 loss=1.177507	 lr=0.000100
Epoch[096] Batch [1649]/[3759]	Speed: 67.702981 samples/sec	 accuracy=70.751894	 loss=1.178948	 lr=0.000100
Epoch[096] Batch [1699]/[3759]	Speed: 72.207753 samples/sec	 accuracy=70.739890	 loss=1.179013	 lr=0.000100
Epoch[096] Batch [1749]/[3759]	Speed: 69.389554 samples/sec	 accuracy=70.756250	 loss=1.178191	 lr=0.000100
Epoch[096] Batch [1799]/[3759]	Speed: 70.716964 samples/sec	 accuracy=70.727431	 loss=1.178814	 lr=0.000100
Epoch[096] Batch [1849]/[3759]	Speed: 69.426074 samples/sec	 accuracy=70.766892	 loss=1.176613	 lr=0.000100
Epoch[096] Batch [1899]/[3759]	Speed: 69.265402 samples/sec	 accuracy=70.725329	 loss=1.177258	 lr=0.000100
Epoch[096] Batch [1949]/[3759]	Speed: 70.921692 samples/sec	 accuracy=70.717949	 loss=1.178163	 lr=0.000100
Epoch[096] Batch [1999]/[3759]	Speed: 69.875901 samples/sec	 accuracy=70.708594	 loss=1.177766	 lr=0.000100
Epoch[096] Batch [2049]/[3759]	Speed: 69.819970 samples/sec	 accuracy=70.721799	 loss=1.177742	 lr=0.000100
Epoch[096] Batch [2099]/[3759]	Speed: 68.581174 samples/sec	 accuracy=70.704613	 loss=1.177958	 lr=0.000100
Epoch[096] Batch [2149]/[3759]	Speed: 70.536175 samples/sec	 accuracy=70.723837	 loss=1.177584	 lr=0.000100
Epoch[096] Batch [2199]/[3759]	Speed: 68.212726 samples/sec	 accuracy=70.718750	 loss=1.177865	 lr=0.000100
Epoch[096] Batch [2249]/[3759]	Speed: 70.010785 samples/sec	 accuracy=70.736111	 loss=1.177672	 lr=0.000100
Epoch[096] Batch [2299]/[3759]	Speed: 70.172376 samples/sec	 accuracy=70.707201	 loss=1.178420	 lr=0.000100
Epoch[096] Batch [2349]/[3759]	Speed: 70.476365 samples/sec	 accuracy=70.696809	 loss=1.179099	 lr=0.000100
Epoch[096] Batch [2399]/[3759]	Speed: 68.175259 samples/sec	 accuracy=70.691406	 loss=1.179299	 lr=0.000100
Epoch[096] Batch [2449]/[3759]	Speed: 71.336466 samples/sec	 accuracy=70.698342	 loss=1.179009	 lr=0.000100
Epoch[096] Batch [2499]/[3759]	Speed: 69.092155 samples/sec	 accuracy=70.716250	 loss=1.178004	 lr=0.000100
Epoch[096] Batch [2549]/[3759]	Speed: 71.181406 samples/sec	 accuracy=70.710172	 loss=1.177843	 lr=0.000100
Epoch[096] Batch [2599]/[3759]	Speed: 68.929885 samples/sec	 accuracy=70.713341	 loss=1.177679	 lr=0.000100
Epoch[096] Batch [2649]/[3759]	Speed: 68.952034 samples/sec	 accuracy=70.719340	 loss=1.177009	 lr=0.000100
Epoch[096] Batch [2699]/[3759]	Speed: 68.609272 samples/sec	 accuracy=70.695602	 loss=1.177896	 lr=0.000100
Epoch[096] Batch [2749]/[3759]	Speed: 71.025590 samples/sec	 accuracy=70.688068	 loss=1.177615	 lr=0.000100
Epoch[096] Batch [2799]/[3759]	Speed: 70.192186 samples/sec	 accuracy=70.695312	 loss=1.177165	 lr=0.000100
Epoch[096] Batch [2849]/[3759]	Speed: 70.282707 samples/sec	 accuracy=70.696272	 loss=1.177136	 lr=0.000100
Epoch[096] Batch [2899]/[3759]	Speed: 69.120867 samples/sec	 accuracy=70.684267	 loss=1.177873	 lr=0.000100
Epoch[096] Batch [2949]/[3759]	Speed: 70.148538 samples/sec	 accuracy=70.673199	 loss=1.178171	 lr=0.000100
Epoch[096] Batch [2999]/[3759]	Speed: 69.939322 samples/sec	 accuracy=70.672396	 loss=1.177911	 lr=0.000100
Epoch[096] Batch [3049]/[3759]	Speed: 69.542303 samples/sec	 accuracy=70.664959	 loss=1.177780	 lr=0.000100
Epoch[096] Batch [3099]/[3759]	Speed: 70.152935 samples/sec	 accuracy=70.685988	 loss=1.177443	 lr=0.000100
Epoch[096] Batch [3149]/[3759]	Speed: 69.918810 samples/sec	 accuracy=70.696429	 loss=1.177285	 lr=0.000100
Epoch[096] Batch [3199]/[3759]	Speed: 68.174286 samples/sec	 accuracy=70.689453	 loss=1.177638	 lr=0.000100
Epoch[096] Batch [3249]/[3759]	Speed: 70.986721 samples/sec	 accuracy=70.676923	 loss=1.177950	 lr=0.000100
Epoch[096] Batch [3299]/[3759]	Speed: 70.367996 samples/sec	 accuracy=70.680871	 loss=1.178134	 lr=0.000100
Epoch[096] Batch [3349]/[3759]	Speed: 69.280456 samples/sec	 accuracy=70.668377	 loss=1.178376	 lr=0.000100
Epoch[096] Batch [3399]/[3759]	Speed: 69.017613 samples/sec	 accuracy=70.670037	 loss=1.178628	 lr=0.000100
Epoch[096] Batch [3449]/[3759]	Speed: 70.101074 samples/sec	 accuracy=70.655344	 loss=1.179038	 lr=0.000100
Epoch[096] Batch [3499]/[3759]	Speed: 70.206303 samples/sec	 accuracy=70.668750	 loss=1.179012	 lr=0.000100
Epoch[096] Batch [3549]/[3759]	Speed: 68.641104 samples/sec	 accuracy=70.669454	 loss=1.178712	 lr=0.000100
Epoch[096] Batch [3599]/[3759]	Speed: 70.666333 samples/sec	 accuracy=70.664062	 loss=1.179134	 lr=0.000100
Epoch[096] Batch [3649]/[3759]	Speed: 70.891041 samples/sec	 accuracy=70.682363	 loss=1.178822	 lr=0.000100
Epoch[096] Batch [3699]/[3759]	Speed: 68.894072 samples/sec	 accuracy=70.679899	 loss=1.178743	 lr=0.000100
Epoch[096] Batch [3749]/[3759]	Speed: 78.766595 samples/sec	 accuracy=70.660833	 loss=1.179590	 lr=0.000100
Batch [0049]/[0303]: acc-top1=62.468750 acc-top5=84.093750
Batch [0099]/[0303]: acc-top1=62.875000 acc-top5=84.281250
Batch [0149]/[0303]: acc-top1=63.958333 acc-top5=84.604167
Batch [0199]/[0303]: acc-top1=64.437500 acc-top5=84.984375
Batch [0249]/[0303]: acc-top1=64.643750 acc-top5=85.225000
Batch [0299]/[0303]: acc-top1=64.510417 acc-top5=85.104167
[Epoch 096] training: accuracy=70.661662	 loss=1.179604
[Epoch 096] speed: 69 samples/sec	time cost: 3746.118063
[Epoch 096] validation: acc-top1=64.490512 acc-top5=85.117574 loss=1.550950
Epoch[097] Batch [0049]/[3760]	Speed: 46.849037 samples/sec	 accuracy=69.406250	 loss=1.167900	 lr=0.000100
Epoch[097] Batch [0099]/[3760]	Speed: 71.354366 samples/sec	 accuracy=70.359375	 loss=1.172348	 lr=0.000100
Epoch[097] Batch [0149]/[3760]	Speed: 68.536143 samples/sec	 accuracy=70.437500	 loss=1.178306	 lr=0.000100
Epoch[097] Batch [0199]/[3760]	Speed: 68.882257 samples/sec	 accuracy=70.429688	 loss=1.179292	 lr=0.000100
Epoch[097] Batch [0249]/[3760]	Speed: 68.806292 samples/sec	 accuracy=70.350000	 loss=1.181887	 lr=0.000100
Epoch[097] Batch [0299]/[3760]	Speed: 70.923924 samples/sec	 accuracy=70.416667	 loss=1.176891	 lr=0.000100
Epoch[097] Batch [0349]/[3760]	Speed: 70.668348 samples/sec	 accuracy=70.348214	 loss=1.185175	 lr=0.000100
Epoch[097] Batch [0399]/[3760]	Speed: 69.887816 samples/sec	 accuracy=70.386719	 loss=1.184392	 lr=0.000100
Epoch[097] Batch [0449]/[3760]	Speed: 69.800609 samples/sec	 accuracy=70.531250	 loss=1.181411	 lr=0.000100
Epoch[097] Batch [0499]/[3760]	Speed: 69.488263 samples/sec	 accuracy=70.418750	 loss=1.184577	 lr=0.000100
Epoch[097] Batch [0549]/[3760]	Speed: 68.208141 samples/sec	 accuracy=70.366477	 loss=1.184186	 lr=0.000100
Epoch[097] Batch [0599]/[3760]	Speed: 70.196782 samples/sec	 accuracy=70.372396	 loss=1.182850	 lr=0.000100
Epoch[097] Batch [0649]/[3760]	Speed: 69.595699 samples/sec	 accuracy=70.425481	 loss=1.181602	 lr=0.000100
Epoch[097] Batch [0699]/[3760]	Speed: 69.864704 samples/sec	 accuracy=70.466518	 loss=1.180115	 lr=0.000100
Epoch[097] Batch [0749]/[3760]	Speed: 68.572163 samples/sec	 accuracy=70.545833	 loss=1.176967	 lr=0.000100
Epoch[097] Batch [0799]/[3760]	Speed: 70.607851 samples/sec	 accuracy=70.625000	 loss=1.176118	 lr=0.000100
Epoch[097] Batch [0849]/[3760]	Speed: 68.891265 samples/sec	 accuracy=70.628676	 loss=1.177131	 lr=0.000100
Epoch[097] Batch [0899]/[3760]	Speed: 70.356726 samples/sec	 accuracy=70.684028	 loss=1.175744	 lr=0.000100
Epoch[097] Batch [0949]/[3760]	Speed: 69.812998 samples/sec	 accuracy=70.672697	 loss=1.176415	 lr=0.000100
Epoch[097] Batch [0999]/[3760]	Speed: 68.315135 samples/sec	 accuracy=70.653125	 loss=1.175961	 lr=0.000100
Epoch[097] Batch [1049]/[3760]	Speed: 69.899939 samples/sec	 accuracy=70.659226	 loss=1.174877	 lr=0.000100
Epoch[097] Batch [1099]/[3760]	Speed: 70.901882 samples/sec	 accuracy=70.711648	 loss=1.172146	 lr=0.000100
Epoch[097] Batch [1149]/[3760]	Speed: 68.987104 samples/sec	 accuracy=70.703804	 loss=1.173224	 lr=0.000100
Epoch[097] Batch [1199]/[3760]	Speed: 70.400135 samples/sec	 accuracy=70.677083	 loss=1.174031	 lr=0.000100
Epoch[097] Batch [1249]/[3760]	Speed: 68.208177 samples/sec	 accuracy=70.688750	 loss=1.173530	 lr=0.000100
Epoch[097] Batch [1299]/[3760]	Speed: 70.116845 samples/sec	 accuracy=70.697115	 loss=1.172899	 lr=0.000100
Epoch[097] Batch [1349]/[3760]	Speed: 70.307696 samples/sec	 accuracy=70.697917	 loss=1.172848	 lr=0.000100
Epoch[097] Batch [1399]/[3760]	Speed: 69.465556 samples/sec	 accuracy=70.710938	 loss=1.172876	 lr=0.000100
Epoch[097] Batch [1449]/[3760]	Speed: 68.482439 samples/sec	 accuracy=70.745690	 loss=1.171260	 lr=0.000100
Epoch[097] Batch [1499]/[3760]	Speed: 68.432447 samples/sec	 accuracy=70.743750	 loss=1.172313	 lr=0.000100
Epoch[097] Batch [1549]/[3760]	Speed: 71.848537 samples/sec	 accuracy=70.738911	 loss=1.172594	 lr=0.000100
Epoch[097] Batch [1599]/[3760]	Speed: 70.306623 samples/sec	 accuracy=70.793945	 loss=1.171464	 lr=0.000100
Epoch[097] Batch [1649]/[3760]	Speed: 69.074634 samples/sec	 accuracy=70.827652	 loss=1.170743	 lr=0.000100
Epoch[097] Batch [1699]/[3760]	Speed: 69.422982 samples/sec	 accuracy=70.854779	 loss=1.170866	 lr=0.000100
Epoch[097] Batch [1749]/[3760]	Speed: 70.100651 samples/sec	 accuracy=70.868750	 loss=1.170783	 lr=0.000100
Epoch[097] Batch [1799]/[3760]	Speed: 69.598142 samples/sec	 accuracy=70.855903	 loss=1.170736	 lr=0.000100
Epoch[097] Batch [1849]/[3760]	Speed: 69.111127 samples/sec	 accuracy=70.822635	 loss=1.172140	 lr=0.000100
Epoch[097] Batch [1899]/[3760]	Speed: 69.444371 samples/sec	 accuracy=70.810033	 loss=1.173362	 lr=0.000100
Epoch[097] Batch [1949]/[3760]	Speed: 70.455436 samples/sec	 accuracy=70.795673	 loss=1.173487	 lr=0.000100
Epoch[097] Batch [1999]/[3760]	Speed: 69.202242 samples/sec	 accuracy=70.839063	 loss=1.172098	 lr=0.000100
Epoch[097] Batch [2049]/[3760]	Speed: 68.537489 samples/sec	 accuracy=70.839177	 loss=1.172019	 lr=0.000100
Epoch[097] Batch [2099]/[3760]	Speed: 70.828994 samples/sec	 accuracy=70.837798	 loss=1.171799	 lr=0.000100
Epoch[097] Batch [2149]/[3760]	Speed: 69.848417 samples/sec	 accuracy=70.834302	 loss=1.172325	 lr=0.000100
Epoch[097] Batch [2199]/[3760]	Speed: 70.081324 samples/sec	 accuracy=70.830966	 loss=1.172637	 lr=0.000100
Epoch[097] Batch [2249]/[3760]	Speed: 68.632141 samples/sec	 accuracy=70.797222	 loss=1.174378	 lr=0.000100
Epoch[097] Batch [2299]/[3760]	Speed: 69.287174 samples/sec	 accuracy=70.764946	 loss=1.175924	 lr=0.000100
Epoch[097] Batch [2349]/[3760]	Speed: 69.215700 samples/sec	 accuracy=70.728723	 loss=1.177047	 lr=0.000100
Epoch[097] Batch [2399]/[3760]	Speed: 68.952720 samples/sec	 accuracy=70.727865	 loss=1.177695	 lr=0.000100
Epoch[097] Batch [2449]/[3760]	Speed: 71.317228 samples/sec	 accuracy=70.720026	 loss=1.178159	 lr=0.000100
Epoch[097] Batch [2499]/[3760]	Speed: 68.664122 samples/sec	 accuracy=70.728750	 loss=1.178152	 lr=0.000100
Epoch[097] Batch [2549]/[3760]	Speed: 69.648135 samples/sec	 accuracy=70.719975	 loss=1.178641	 lr=0.000100
Epoch[097] Batch [2599]/[3760]	Speed: 70.019561 samples/sec	 accuracy=70.725361	 loss=1.178658	 lr=0.000100
Epoch[097] Batch [2649]/[3760]	Speed: 68.213280 samples/sec	 accuracy=70.734670	 loss=1.178330	 lr=0.000100
Epoch[097] Batch [2699]/[3760]	Speed: 70.081717 samples/sec	 accuracy=70.743056	 loss=1.178469	 lr=0.000100
Epoch[097] Batch [2749]/[3760]	Speed: 69.710619 samples/sec	 accuracy=70.738068	 loss=1.179242	 lr=0.000100
Epoch[097] Batch [2799]/[3760]	Speed: 70.399912 samples/sec	 accuracy=70.743304	 loss=1.179470	 lr=0.000100
Epoch[097] Batch [2849]/[3760]	Speed: 68.364080 samples/sec	 accuracy=70.757675	 loss=1.178873	 lr=0.000100
Epoch[097] Batch [2899]/[3760]	Speed: 70.907400 samples/sec	 accuracy=70.756466	 loss=1.179762	 lr=0.000100
Epoch[097] Batch [2949]/[3760]	Speed: 66.509406 samples/sec	 accuracy=70.740466	 loss=1.179698	 lr=0.000100
Epoch[097] Batch [2999]/[3760]	Speed: 72.771871 samples/sec	 accuracy=70.764583	 loss=1.178957	 lr=0.000100
Epoch[097] Batch [3049]/[3760]	Speed: 69.000033 samples/sec	 accuracy=70.754098	 loss=1.179458	 lr=0.000100
Epoch[097] Batch [3099]/[3760]	Speed: 69.407953 samples/sec	 accuracy=70.746472	 loss=1.179554	 lr=0.000100
Epoch[097] Batch [3149]/[3760]	Speed: 69.719147 samples/sec	 accuracy=70.755456	 loss=1.179552	 lr=0.000100
Epoch[097] Batch [3199]/[3760]	Speed: 69.934588 samples/sec	 accuracy=70.761230	 loss=1.179421	 lr=0.000100
Epoch[097] Batch [3249]/[3760]	Speed: 65.956146 samples/sec	 accuracy=70.764423	 loss=1.179012	 lr=0.000100
Epoch[097] Batch [3299]/[3760]	Speed: 73.779681 samples/sec	 accuracy=70.763258	 loss=1.178948	 lr=0.000100
Epoch[097] Batch [3349]/[3760]	Speed: 69.632625 samples/sec	 accuracy=70.772388	 loss=1.178487	 lr=0.000100
Epoch[097] Batch [3399]/[3760]	Speed: 70.236849 samples/sec	 accuracy=70.762408	 loss=1.179206	 lr=0.000100
Epoch[097] Batch [3449]/[3760]	Speed: 70.169332 samples/sec	 accuracy=70.794837	 loss=1.178377	 lr=0.000100
Epoch[097] Batch [3499]/[3760]	Speed: 69.367682 samples/sec	 accuracy=70.796875	 loss=1.177797	 lr=0.000100
Epoch[097] Batch [3549]/[3760]	Speed: 69.342021 samples/sec	 accuracy=70.795335	 loss=1.178423	 lr=0.000100
Epoch[097] Batch [3599]/[3760]	Speed: 70.228591 samples/sec	 accuracy=70.781684	 loss=1.178545	 lr=0.000100
Epoch[097] Batch [3649]/[3760]	Speed: 69.216348 samples/sec	 accuracy=70.786387	 loss=1.178597	 lr=0.000100
Epoch[097] Batch [3699]/[3760]	Speed: 70.296003 samples/sec	 accuracy=70.788007	 loss=1.178504	 lr=0.000100
Epoch[097] Batch [3749]/[3760]	Speed: 77.739984 samples/sec	 accuracy=70.786667	 loss=1.178580	 lr=0.000100
Batch [0049]/[0303]: acc-top1=63.062500 acc-top5=84.281250
Batch [0099]/[0303]: acc-top1=62.968750 acc-top5=84.390625
Batch [0149]/[0303]: acc-top1=63.989583 acc-top5=84.718750
Batch [0199]/[0303]: acc-top1=64.500000 acc-top5=85.101562
Batch [0249]/[0303]: acc-top1=64.675000 acc-top5=85.300000
Batch [0299]/[0303]: acc-top1=64.479167 acc-top5=85.140625
[Epoch 097] training: accuracy=70.782497	 loss=1.178596
[Epoch 097] speed: 69 samples/sec	time cost: 3755.321495
[Epoch 097] validation: acc-top1=64.475041 acc-top5=85.158828 loss=1.556889
Epoch[098] Batch [0049]/[3759]	Speed: 47.790467 samples/sec	 accuracy=70.906250	 loss=1.155558	 lr=0.000100
Epoch[098] Batch [0099]/[3759]	Speed: 70.980809 samples/sec	 accuracy=70.796875	 loss=1.180953	 lr=0.000100
Epoch[098] Batch [0149]/[3759]	Speed: 67.777359 samples/sec	 accuracy=70.697917	 loss=1.184069	 lr=0.000100
Epoch[098] Batch [0199]/[3759]	Speed: 69.457746 samples/sec	 accuracy=70.398438	 loss=1.192494	 lr=0.000100
Epoch[098] Batch [0249]/[3759]	Speed: 69.879209 samples/sec	 accuracy=70.487500	 loss=1.190312	 lr=0.000100
Epoch[098] Batch [0299]/[3759]	Speed: 69.801538 samples/sec	 accuracy=70.796875	 loss=1.181760	 lr=0.000100
Epoch[098] Batch [0349]/[3759]	Speed: 65.108364 samples/sec	 accuracy=70.955357	 loss=1.173241	 lr=0.000100
Epoch[098] Batch [0399]/[3759]	Speed: 75.271198 samples/sec	 accuracy=70.812500	 loss=1.176189	 lr=0.000100
Epoch[098] Batch [0449]/[3759]	Speed: 69.334150 samples/sec	 accuracy=70.881944	 loss=1.175466	 lr=0.000100
Epoch[098] Batch [0499]/[3759]	Speed: 68.914281 samples/sec	 accuracy=71.003125	 loss=1.172635	 lr=0.000100
Epoch[098] Batch [0549]/[3759]	Speed: 70.862351 samples/sec	 accuracy=70.931818	 loss=1.173280	 lr=0.000100
Epoch[098] Batch [0599]/[3759]	Speed: 69.656008 samples/sec	 accuracy=70.882812	 loss=1.171743	 lr=0.000100
Epoch[098] Batch [0649]/[3759]	Speed: 69.465558 samples/sec	 accuracy=70.889423	 loss=1.172585	 lr=0.000100
Epoch[098] Batch [0699]/[3759]	Speed: 69.348621 samples/sec	 accuracy=71.022321	 loss=1.167705	 lr=0.000100
Epoch[098] Batch [0749]/[3759]	Speed: 66.982217 samples/sec	 accuracy=70.995833	 loss=1.166870	 lr=0.000100
Epoch[098] Batch [0799]/[3759]	Speed: 71.073268 samples/sec	 accuracy=70.964844	 loss=1.169410	 lr=0.000100
Epoch[098] Batch [0849]/[3759]	Speed: 70.025608 samples/sec	 accuracy=71.005515	 loss=1.168157	 lr=0.000100
Epoch[098] Batch [0899]/[3759]	Speed: 70.823915 samples/sec	 accuracy=71.026042	 loss=1.169477	 lr=0.000100
Epoch[098] Batch [0949]/[3759]	Speed: 68.475891 samples/sec	 accuracy=70.993421	 loss=1.170151	 lr=0.000100
Epoch[098] Batch [0999]/[3759]	Speed: 71.478719 samples/sec	 accuracy=70.978125	 loss=1.169827	 lr=0.000100
Epoch[098] Batch [1049]/[3759]	Speed: 68.582892 samples/sec	 accuracy=71.002976	 loss=1.169840	 lr=0.000100
Epoch[098] Batch [1099]/[3759]	Speed: 70.125218 samples/sec	 accuracy=71.002841	 loss=1.170194	 lr=0.000100
Epoch[098] Batch [1149]/[3759]	Speed: 69.402014 samples/sec	 accuracy=70.975543	 loss=1.170307	 lr=0.000100
Epoch[098] Batch [1199]/[3759]	Speed: 70.101525 samples/sec	 accuracy=70.993490	 loss=1.170757	 lr=0.000100
Epoch[098] Batch [1249]/[3759]	Speed: 67.796248 samples/sec	 accuracy=70.981250	 loss=1.170828	 lr=0.000100
Epoch[098] Batch [1299]/[3759]	Speed: 67.692367 samples/sec	 accuracy=70.968750	 loss=1.172332	 lr=0.000100
Epoch[098] Batch [1349]/[3759]	Speed: 72.558705 samples/sec	 accuracy=70.952546	 loss=1.172872	 lr=0.000100
Epoch[098] Batch [1399]/[3759]	Speed: 67.412132 samples/sec	 accuracy=70.943080	 loss=1.172642	 lr=0.000100
Epoch[098] Batch [1449]/[3759]	Speed: 69.920139 samples/sec	 accuracy=70.987069	 loss=1.172415	 lr=0.000100
Epoch[098] Batch [1499]/[3759]	Speed: 70.263892 samples/sec	 accuracy=70.973958	 loss=1.173224	 lr=0.000100
Epoch[098] Batch [1549]/[3759]	Speed: 71.177029 samples/sec	 accuracy=70.938508	 loss=1.174647	 lr=0.000100
Epoch[098] Batch [1599]/[3759]	Speed: 69.605279 samples/sec	 accuracy=70.886719	 loss=1.175704	 lr=0.000100
Epoch[098] Batch [1649]/[3759]	Speed: 69.952568 samples/sec	 accuracy=70.875000	 loss=1.176268	 lr=0.000100
Epoch[098] Batch [1699]/[3759]	Speed: 70.336053 samples/sec	 accuracy=70.897978	 loss=1.175024	 lr=0.000100
Epoch[098] Batch [1749]/[3759]	Speed: 68.599154 samples/sec	 accuracy=70.905357	 loss=1.175309	 lr=0.000100
Epoch[098] Batch [1799]/[3759]	Speed: 69.582546 samples/sec	 accuracy=70.913194	 loss=1.175608	 lr=0.000100
Epoch[098] Batch [1849]/[3759]	Speed: 68.712376 samples/sec	 accuracy=70.928209	 loss=1.174449	 lr=0.000100
Epoch[098] Batch [1899]/[3759]	Speed: 70.609915 samples/sec	 accuracy=70.933388	 loss=1.175492	 lr=0.000100
Epoch[098] Batch [1949]/[3759]	Speed: 70.413363 samples/sec	 accuracy=70.931090	 loss=1.174826	 lr=0.000100
Epoch[098] Batch [1999]/[3759]	Speed: 69.466222 samples/sec	 accuracy=70.932812	 loss=1.174469	 lr=0.000100
Epoch[098] Batch [2049]/[3759]	Speed: 68.757692 samples/sec	 accuracy=70.899390	 loss=1.175727	 lr=0.000100
Epoch[098] Batch [2099]/[3759]	Speed: 69.681737 samples/sec	 accuracy=70.883929	 loss=1.176097	 lr=0.000100
Epoch[098] Batch [2149]/[3759]	Speed: 70.076229 samples/sec	 accuracy=70.864099	 loss=1.176908	 lr=0.000100
Epoch[098] Batch [2199]/[3759]	Speed: 69.956303 samples/sec	 accuracy=70.867188	 loss=1.177349	 lr=0.000100
Epoch[098] Batch [2249]/[3759]	Speed: 69.914090 samples/sec	 accuracy=70.869444	 loss=1.177948	 lr=0.000100
Epoch[098] Batch [2299]/[3759]	Speed: 69.604935 samples/sec	 accuracy=70.836957	 loss=1.178747	 lr=0.000100
Epoch[098] Batch [2349]/[3759]	Speed: 69.243869 samples/sec	 accuracy=70.814495	 loss=1.178685	 lr=0.000100
Epoch[098] Batch [2399]/[3759]	Speed: 70.500273 samples/sec	 accuracy=70.819010	 loss=1.178293	 lr=0.000100
Epoch[098] Batch [2449]/[3759]	Speed: 70.474123 samples/sec	 accuracy=70.795281	 loss=1.179333	 lr=0.000100
Epoch[098] Batch [2499]/[3759]	Speed: 68.217870 samples/sec	 accuracy=70.795000	 loss=1.179587	 lr=0.000100
Epoch[098] Batch [2549]/[3759]	Speed: 71.432882 samples/sec	 accuracy=70.775735	 loss=1.179969	 lr=0.000100
Epoch[098] Batch [2599]/[3759]	Speed: 69.726303 samples/sec	 accuracy=70.782452	 loss=1.179244	 lr=0.000100
Epoch[098] Batch [2649]/[3759]	Speed: 69.134986 samples/sec	 accuracy=70.764151	 loss=1.179444	 lr=0.000100
Epoch[098] Batch [2699]/[3759]	Speed: 69.015363 samples/sec	 accuracy=70.755787	 loss=1.179155	 lr=0.000100
Epoch[098] Batch [2749]/[3759]	Speed: 70.832003 samples/sec	 accuracy=70.755114	 loss=1.179205	 lr=0.000100
Epoch[098] Batch [2799]/[3759]	Speed: 70.097154 samples/sec	 accuracy=70.763393	 loss=1.179811	 lr=0.000100
Epoch[098] Batch [2849]/[3759]	Speed: 68.480445 samples/sec	 accuracy=70.776316	 loss=1.179357	 lr=0.000100
Epoch[098] Batch [2899]/[3759]	Speed: 69.468202 samples/sec	 accuracy=70.772629	 loss=1.179486	 lr=0.000100
Epoch[098] Batch [2949]/[3759]	Speed: 70.912205 samples/sec	 accuracy=70.774364	 loss=1.179674	 lr=0.000100
Epoch[098] Batch [2999]/[3759]	Speed: 69.345503 samples/sec	 accuracy=70.778125	 loss=1.178988	 lr=0.000100
Epoch[098] Batch [3049]/[3759]	Speed: 70.062169 samples/sec	 accuracy=70.764344	 loss=1.179777	 lr=0.000100
Epoch[098] Batch [3099]/[3759]	Speed: 69.196418 samples/sec	 accuracy=70.756048	 loss=1.180286	 lr=0.000100
Epoch[098] Batch [3149]/[3759]	Speed: 69.765080 samples/sec	 accuracy=70.757440	 loss=1.179875	 lr=0.000100
Epoch[098] Batch [3199]/[3759]	Speed: 71.049419 samples/sec	 accuracy=70.766602	 loss=1.179393	 lr=0.000100
Epoch[098] Batch [3249]/[3759]	Speed: 67.543857 samples/sec	 accuracy=70.775000	 loss=1.179203	 lr=0.000100
Epoch[098] Batch [3299]/[3759]	Speed: 71.325347 samples/sec	 accuracy=70.776042	 loss=1.179191	 lr=0.000100
Epoch[098] Batch [3349]/[3759]	Speed: 69.351256 samples/sec	 accuracy=70.775653	 loss=1.179389	 lr=0.000100
Epoch[098] Batch [3399]/[3759]	Speed: 69.747170 samples/sec	 accuracy=70.799632	 loss=1.178653	 lr=0.000100
Epoch[098] Batch [3449]/[3759]	Speed: 69.582216 samples/sec	 accuracy=70.802536	 loss=1.178386	 lr=0.000100
Epoch[098] Batch [3499]/[3759]	Speed: 70.123522 samples/sec	 accuracy=70.791071	 loss=1.179056	 lr=0.000100
Epoch[098] Batch [3549]/[3759]	Speed: 68.505909 samples/sec	 accuracy=70.794014	 loss=1.178909	 lr=0.000100
Epoch[098] Batch [3599]/[3759]	Speed: 70.245796 samples/sec	 accuracy=70.802083	 loss=1.178191	 lr=0.000100
Epoch[098] Batch [3649]/[3759]	Speed: 69.426278 samples/sec	 accuracy=70.808647	 loss=1.177799	 lr=0.000100
Epoch[098] Batch [3699]/[3759]	Speed: 70.474606 samples/sec	 accuracy=70.799409	 loss=1.177840	 lr=0.000100
Epoch[098] Batch [3749]/[3759]	Speed: 77.279827 samples/sec	 accuracy=70.796250	 loss=1.178206	 lr=0.000100
Batch [0049]/[0303]: acc-top1=62.375000 acc-top5=84.406250
Batch [0099]/[0303]: acc-top1=62.781250 acc-top5=84.265625
Batch [0149]/[0303]: acc-top1=63.718750 acc-top5=84.708333
Batch [0199]/[0303]: acc-top1=64.218750 acc-top5=85.101562
Batch [0249]/[0303]: acc-top1=64.443750 acc-top5=85.368750
Batch [0299]/[0303]: acc-top1=64.291667 acc-top5=85.234375
[Epoch 098] training: accuracy=70.794676	 loss=1.177990
[Epoch 098] speed: 69 samples/sec	time cost: 3751.830361
[Epoch 098] validation: acc-top1=64.258457 acc-top5=85.241337 loss=1.556249
Epoch[099] Batch [0049]/[3760]	Speed: 48.972603 samples/sec	 accuracy=72.218750	 loss=1.158093	 lr=0.000100
Epoch[099] Batch [0099]/[3760]	Speed: 67.697660 samples/sec	 accuracy=71.671875	 loss=1.166579	 lr=0.000100
Epoch[099] Batch [0149]/[3760]	Speed: 69.515633 samples/sec	 accuracy=71.385417	 loss=1.170189	 lr=0.000100
Epoch[099] Batch [0199]/[3760]	Speed: 70.552475 samples/sec	 accuracy=71.304688	 loss=1.170043	 lr=0.000100
Epoch[099] Batch [0249]/[3760]	Speed: 69.465067 samples/sec	 accuracy=71.350000	 loss=1.169573	 lr=0.000100
Epoch[099] Batch [0299]/[3760]	Speed: 69.400823 samples/sec	 accuracy=71.223958	 loss=1.173008	 lr=0.000100
Epoch[099] Batch [0349]/[3760]	Speed: 69.324118 samples/sec	 accuracy=71.200893	 loss=1.174038	 lr=0.000100
Epoch[099] Batch [0399]/[3760]	Speed: 70.707780 samples/sec	 accuracy=71.339844	 loss=1.169663	 lr=0.000100
Epoch[099] Batch [0449]/[3760]	Speed: 70.351325 samples/sec	 accuracy=71.260417	 loss=1.176069	 lr=0.000100
Epoch[099] Batch [0499]/[3760]	Speed: 70.684142 samples/sec	 accuracy=71.184375	 loss=1.176826	 lr=0.000100
Epoch[099] Batch [0549]/[3760]	Speed: 68.460006 samples/sec	 accuracy=71.139205	 loss=1.177747	 lr=0.000100
Epoch[099] Batch [0599]/[3760]	Speed: 69.941710 samples/sec	 accuracy=71.119792	 loss=1.178546	 lr=0.000100
Epoch[099] Batch [0649]/[3760]	Speed: 70.777661 samples/sec	 accuracy=71.016827	 loss=1.180413	 lr=0.000100
Epoch[099] Batch [0699]/[3760]	Speed: 69.750408 samples/sec	 accuracy=71.024554	 loss=1.179399	 lr=0.000100
Epoch[099] Batch [0749]/[3760]	Speed: 69.908206 samples/sec	 accuracy=71.035417	 loss=1.179357	 lr=0.000100
Epoch[099] Batch [0799]/[3760]	Speed: 70.343440 samples/sec	 accuracy=71.011719	 loss=1.180387	 lr=0.000100
Epoch[099] Batch [0849]/[3760]	Speed: 68.975496 samples/sec	 accuracy=70.977941	 loss=1.182064	 lr=0.000100
Epoch[099] Batch [0899]/[3760]	Speed: 70.247996 samples/sec	 accuracy=70.949653	 loss=1.184011	 lr=0.000100
Epoch[099] Batch [0949]/[3760]	Speed: 68.581925 samples/sec	 accuracy=71.034539	 loss=1.182517	 lr=0.000100
Epoch[099] Batch [0999]/[3760]	Speed: 68.959818 samples/sec	 accuracy=70.943750	 loss=1.183747	 lr=0.000100
Epoch[099] Batch [1049]/[3760]	Speed: 70.051018 samples/sec	 accuracy=71.063988	 loss=1.179698	 lr=0.000100
Epoch[099] Batch [1099]/[3760]	Speed: 71.005493 samples/sec	 accuracy=71.044034	 loss=1.179654	 lr=0.000100
Epoch[099] Batch [1149]/[3760]	Speed: 69.030760 samples/sec	 accuracy=70.997283	 loss=1.180285	 lr=0.000100
Epoch[099] Batch [1199]/[3760]	Speed: 70.188042 samples/sec	 accuracy=70.971354	 loss=1.181851	 lr=0.000100
Epoch[099] Batch [1249]/[3760]	Speed: 68.900125 samples/sec	 accuracy=70.912500	 loss=1.183484	 lr=0.000100
Epoch[099] Batch [1299]/[3760]	Speed: 70.513425 samples/sec	 accuracy=70.867788	 loss=1.184013	 lr=0.000100
Epoch[099] Batch [1349]/[3760]	Speed: 69.956278 samples/sec	 accuracy=70.888889	 loss=1.181544	 lr=0.000100
Epoch[099] Batch [1399]/[3760]	Speed: 70.479379 samples/sec	 accuracy=70.864955	 loss=1.182597	 lr=0.000100
Epoch[099] Batch [1449]/[3760]	Speed: 69.740096 samples/sec	 accuracy=70.876078	 loss=1.182617	 lr=0.000100
Epoch[099] Batch [1499]/[3760]	Speed: 69.176028 samples/sec	 accuracy=70.870833	 loss=1.183787	 lr=0.000100
Epoch[099] Batch [1549]/[3760]	Speed: 66.438919 samples/sec	 accuracy=70.862903	 loss=1.184591	 lr=0.000100
Epoch[099] Batch [1599]/[3760]	Speed: 72.012104 samples/sec	 accuracy=70.845703	 loss=1.184836	 lr=0.000100
Epoch[099] Batch [1649]/[3760]	Speed: 71.187893 samples/sec	 accuracy=70.863636	 loss=1.183558	 lr=0.000100
Epoch[099] Batch [1699]/[3760]	Speed: 69.617873 samples/sec	 accuracy=70.863051	 loss=1.183785	 lr=0.000100
Epoch[099] Batch [1749]/[3760]	Speed: 69.348977 samples/sec	 accuracy=70.881250	 loss=1.182491	 lr=0.000100
Epoch[099] Batch [1799]/[3760]	Speed: 69.895875 samples/sec	 accuracy=70.901042	 loss=1.180680	 lr=0.000100
Epoch[099] Batch [1849]/[3760]	Speed: 68.388962 samples/sec	 accuracy=70.832770	 loss=1.182220	 lr=0.000100
Epoch[099] Batch [1899]/[3760]	Speed: 70.855704 samples/sec	 accuracy=70.846217	 loss=1.181813	 lr=0.000100
Epoch[099] Batch [1949]/[3760]	Speed: 70.092990 samples/sec	 accuracy=70.855769	 loss=1.181318	 lr=0.000100
Epoch[099] Batch [1999]/[3760]	Speed: 68.034546 samples/sec	 accuracy=70.869531	 loss=1.180534	 lr=0.000100
Epoch[099] Batch [2049]/[3760]	Speed: 70.554379 samples/sec	 accuracy=70.832317	 loss=1.181262	 lr=0.000100
Epoch[099] Batch [2099]/[3760]	Speed: 70.401635 samples/sec	 accuracy=70.816964	 loss=1.181946	 lr=0.000100
Epoch[099] Batch [2149]/[3760]	Speed: 70.044129 samples/sec	 accuracy=70.825581	 loss=1.181529	 lr=0.000100
Epoch[099] Batch [2199]/[3760]	Speed: 68.605722 samples/sec	 accuracy=70.835938	 loss=1.181278	 lr=0.000100
Epoch[099] Batch [2249]/[3760]	Speed: 68.831747 samples/sec	 accuracy=70.829167	 loss=1.181820	 lr=0.000100
Epoch[099] Batch [2299]/[3760]	Speed: 71.228909 samples/sec	 accuracy=70.804348	 loss=1.183028	 lr=0.000100
Epoch[099] Batch [2349]/[3760]	Speed: 68.003881 samples/sec	 accuracy=70.798537	 loss=1.183050	 lr=0.000100
Epoch[099] Batch [2399]/[3760]	Speed: 69.937308 samples/sec	 accuracy=70.835286	 loss=1.181485	 lr=0.000100
Epoch[099] Batch [2449]/[3760]	Speed: 71.314453 samples/sec	 accuracy=70.845663	 loss=1.180747	 lr=0.000100
Epoch[099] Batch [2499]/[3760]	Speed: 70.155485 samples/sec	 accuracy=70.866875	 loss=1.180190	 lr=0.000100
Epoch[099] Batch [2549]/[3760]	Speed: 67.930319 samples/sec	 accuracy=70.880515	 loss=1.179475	 lr=0.000100
Epoch[099] Batch [2599]/[3760]	Speed: 70.200052 samples/sec	 accuracy=70.883413	 loss=1.179828	 lr=0.000100
Epoch[099] Batch [2649]/[3760]	Speed: 70.520903 samples/sec	 accuracy=70.886203	 loss=1.180158	 lr=0.000100
Epoch[099] Batch [2699]/[3760]	Speed: 69.940036 samples/sec	 accuracy=70.886574	 loss=1.179943	 lr=0.000100
Epoch[099] Batch [2749]/[3760]	Speed: 69.918437 samples/sec	 accuracy=70.897159	 loss=1.179599	 lr=0.000100
Epoch[099] Batch [2799]/[3760]	Speed: 68.422960 samples/sec	 accuracy=70.885603	 loss=1.179916	 lr=0.000100
Epoch[099] Batch [2849]/[3760]	Speed: 71.851584 samples/sec	 accuracy=70.884320	 loss=1.179115	 lr=0.000100
Epoch[099] Batch [2899]/[3760]	Speed: 70.249804 samples/sec	 accuracy=70.873922	 loss=1.178896	 lr=0.000100
Epoch[099] Batch [2949]/[3760]	Speed: 69.601765 samples/sec	 accuracy=70.875530	 loss=1.178421	 lr=0.000100
Epoch[099] Batch [2999]/[3760]	Speed: 69.056791 samples/sec	 accuracy=70.897917	 loss=1.177791	 lr=0.000100
Epoch[099] Batch [3049]/[3760]	Speed: 70.085419 samples/sec	 accuracy=70.884734	 loss=1.177994	 lr=0.000100
Epoch[099] Batch [3099]/[3760]	Speed: 70.092700 samples/sec	 accuracy=70.893649	 loss=1.177263	 lr=0.000100
Epoch[099] Batch [3149]/[3760]	Speed: 70.336632 samples/sec	 accuracy=70.896329	 loss=1.177055	 lr=0.000100
Epoch[099] Batch [3199]/[3760]	Speed: 68.672643 samples/sec	 accuracy=70.884277	 loss=1.177661	 lr=0.000100
Epoch[099] Batch [3249]/[3760]	Speed: 70.478394 samples/sec	 accuracy=70.890385	 loss=1.177479	 lr=0.000100
Epoch[099] Batch [3299]/[3760]	Speed: 70.557354 samples/sec	 accuracy=70.888258	 loss=1.177627	 lr=0.000100
Epoch[099] Batch [3349]/[3760]	Speed: 70.294028 samples/sec	 accuracy=70.875933	 loss=1.177668	 lr=0.000100
Epoch[099] Batch [3399]/[3760]	Speed: 69.615523 samples/sec	 accuracy=70.868566	 loss=1.177220	 lr=0.000100
Epoch[099] Batch [3449]/[3760]	Speed: 70.427820 samples/sec	 accuracy=70.852808	 loss=1.177539	 lr=0.000100
Epoch[099] Batch [3499]/[3760]	Speed: 68.751349 samples/sec	 accuracy=70.865625	 loss=1.177132	 lr=0.000100
Epoch[099] Batch [3549]/[3760]	Speed: 69.801816 samples/sec	 accuracy=70.853433	 loss=1.177861	 lr=0.000100
Epoch[099] Batch [3599]/[3760]	Speed: 70.492510 samples/sec	 accuracy=70.865017	 loss=1.177454	 lr=0.000100
Epoch[099] Batch [3649]/[3760]	Speed: 68.734642 samples/sec	 accuracy=70.882277	 loss=1.176923	 lr=0.000100
Epoch[099] Batch [3699]/[3760]	Speed: 69.965563 samples/sec	 accuracy=70.888091	 loss=1.177044	 lr=0.000100
Epoch[099] Batch [3749]/[3760]	Speed: 77.782919 samples/sec	 accuracy=70.893333	 loss=1.177005	 lr=0.000100
Batch [0049]/[0303]: acc-top1=62.968750 acc-top5=84.343750
Batch [0099]/[0303]: acc-top1=62.781250 acc-top5=84.500000
Batch [0149]/[0303]: acc-top1=63.812500 acc-top5=84.791667
Batch [0199]/[0303]: acc-top1=64.429688 acc-top5=85.117188
Batch [0249]/[0303]: acc-top1=64.737500 acc-top5=85.293750
Batch [0299]/[0303]: acc-top1=64.562500 acc-top5=85.062500
[Epoch 099] training: accuracy=70.888049	 loss=1.176983
[Epoch 099] speed: 69 samples/sec	time cost: 3746.708582
[Epoch 099] validation: acc-top1=64.547236 acc-top5=85.076320 loss=1.552403