Namespace(accumulate=1, batch_norm=False, batch_size=8, clip_grad=40, crop_ratio=0.875, data_dir='/home/ubuntu/.mxnet/datasets/ucf101/rawframes', dataset='ucf101', dtype='float32', eval=False, hard_weight=0.5, input_5d=False, input_size=224, kvstore=None, label_smoothing=False, last_gamma=False, log_interval=20, logging_file='i3d_resnet50_v1_ucf101_b8_g8_inflate311_f32s2_step_dp8_init001_lr001_epoch50_run1.txt', lr=0.001, lr_decay=0.1, lr_decay_epoch='20,40,50', lr_decay_period=0, lr_mode='step', mixup=False, mixup_alpha=0.2, mixup_off_epoch=0, mode='hybrid', model='i3d_resnet50_v1_ucf101', momentum=0.9, new_height=256, new_length=32, new_step=2, new_width=340, no_wd=False, num_classes=101, num_crop=1, num_epochs=50, 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/ucf101/i3d_resnet50_v1_ucf101_b8_g8_inflate311_f32s2_step_dp8_init001_lr001_epoch50_run1', save_frequency=5, scale_ratios='1.0,0.8', teacher=None, temperature=20, train_list='/home/ubuntu/.mxnet/datasets/ucf101/ucfTrainTestlist/ucf101_train_split_1_rawframes.txt', use_amp=False, use_decord=False, use_gn=False, use_pretrained=False, use_se=False, use_tsn=False, val_data_dir='~/.mxnet/datasets/ucf101/rawframes', val_list='/home/ubuntu/.mxnet/datasets/ucf101/ucfTrainTestlist/ucf101_val_split_1_rawframes.txt', video_loader=False, warmup_epochs=0, warmup_lr=0.0, wd=0.0001) Total batch size is set to 64 on 8 GPUs I3D_ResNetV1( (first_stage): HybridSequential( (0): Conv3D(3 -> 64, kernel_size=(5, 7, 7), stride=(2, 2, 2), padding=(2, 3, 3), bias=False) (1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=64) (2): Activation(relu) (3): MaxPool3D(size=(1, 3, 3), stride=(2, 2, 2), padding=(0, 1, 1), ceil_mode=False, global_pool=False, pool_type=max, layout=NCDHW) ) (pool2): MaxPool3D(size=(2, 1, 1), stride=(2, 1, 1), padding=(0, 0, 0), ceil_mode=False, global_pool=False, pool_type=max, layout=NCDHW) (res_layers): HybridSequential( (0): HybridSequential( (0): Bottleneck( (bottleneck): HybridSequential( (0): Conv3D(64 -> 64, kernel_size=(3, 1, 1), stride=(1, 1, 1), padding=(1, 0, 0), bias=False) (1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=64) (2): Activation(relu) (3): Conv3D(64 -> 64, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1), bias=False) (4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=64) (5): Activation(relu) (6): Conv3D(64 -> 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (7): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=256) ) (conv1): Conv3D(64 -> 64, kernel_size=(3, 1, 1), stride=(1, 1, 1), padding=(1, 0, 0), bias=False) (conv2): Conv3D(64 -> 64, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1), bias=False) (bn1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=64) (bn2): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=64) (conv3): Conv3D(64 -> 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (bn3): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=256) (relu): Activation(relu) (downsample): HybridSequential( (0): Conv3D(64 -> 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=256) ) ) (1): Bottleneck( (bottleneck): HybridSequential( (0): Conv3D(256 -> 64, kernel_size=(3, 1, 1), stride=(1, 1, 1), padding=(1, 0, 0), bias=False) (1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=64) (2): Activation(relu) (3): Conv3D(64 -> 64, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1), bias=False) (4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=64) (5): Activation(relu) (6): Conv3D(64 -> 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (7): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=256) ) (conv1): Conv3D(256 -> 64, kernel_size=(3, 1, 1), stride=(1, 1, 1), padding=(1, 0, 0), bias=False) (conv2): Conv3D(64 -> 64, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1), bias=False) (bn1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=64) (bn2): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=64) (conv3): Conv3D(64 -> 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (bn3): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=256) (relu): Activation(relu) ) (2): Bottleneck( (bottleneck): HybridSequential( (0): Conv3D(256 -> 64, kernel_size=(3, 1, 1), stride=(1, 1, 1), padding=(1, 0, 0), bias=False) (1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=64) (2): Activation(relu) (3): Conv3D(64 -> 64, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1), bias=False) (4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=64) (5): Activation(relu) (6): Conv3D(64 -> 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (7): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=256) ) (conv1): Conv3D(256 -> 64, kernel_size=(3, 1, 1), stride=(1, 1, 1), padding=(1, 0, 0), bias=False) (conv2): Conv3D(64 -> 64, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1), bias=False) (bn1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=64) (bn2): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=64) (conv3): Conv3D(64 -> 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (bn3): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=256) (relu): Activation(relu) ) ) (1): HybridSequential( (0): Bottleneck( (bottleneck): HybridSequential( (0): Conv3D(256 -> 128, kernel_size=(3, 1, 1), stride=(1, 1, 1), padding=(1, 0, 0), bias=False) (1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=128) (2): Activation(relu) (3): Conv3D(128 -> 128, kernel_size=(1, 3, 3), stride=(1, 2, 2), padding=(0, 1, 1), bias=False) (4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=128) (5): Activation(relu) (6): Conv3D(128 -> 512, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (7): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=512) ) (conv1): Conv3D(256 -> 128, kernel_size=(3, 1, 1), stride=(1, 1, 1), padding=(1, 0, 0), bias=False) (conv2): Conv3D(128 -> 128, kernel_size=(1, 3, 3), stride=(1, 2, 2), padding=(0, 1, 1), bias=False) (bn1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=128) (bn2): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=128) (conv3): Conv3D(128 -> 512, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (bn3): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=512) (relu): Activation(relu) (downsample): HybridSequential( (0): Conv3D(256 -> 512, kernel_size=(1, 1, 1), stride=(1, 2, 2), bias=False) (1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=512) ) ) (1): Bottleneck( (bottleneck): HybridSequential( (0): Conv3D(512 -> 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=128) (2): Activation(relu) (3): Conv3D(128 -> 128, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1), bias=False) (4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=128) (5): Activation(relu) (6): Conv3D(128 -> 512, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (7): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=512) ) (conv1): Conv3D(512 -> 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (conv2): Conv3D(128 -> 128, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1), bias=False) (bn1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=128) (bn2): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=128) (conv3): Conv3D(128 -> 512, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (bn3): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=512) (relu): Activation(relu) ) (2): Bottleneck( (bottleneck): HybridSequential( (0): Conv3D(512 -> 128, kernel_size=(3, 1, 1), stride=(1, 1, 1), padding=(1, 0, 0), bias=False) (1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=128) (2): Activation(relu) (3): Conv3D(128 -> 128, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1), bias=False) (4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=128) (5): Activation(relu) (6): Conv3D(128 -> 512, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (7): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=512) ) (conv1): Conv3D(512 -> 128, kernel_size=(3, 1, 1), stride=(1, 1, 1), padding=(1, 0, 0), bias=False) (conv2): Conv3D(128 -> 128, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1), bias=False) (bn1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=128) (bn2): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=128) (conv3): Conv3D(128 -> 512, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (bn3): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=512) (relu): Activation(relu) ) (3): Bottleneck( (bottleneck): HybridSequential( (0): Conv3D(512 -> 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=128) (2): Activation(relu) (3): Conv3D(128 -> 128, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1), bias=False) (4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=128) (5): Activation(relu) (6): Conv3D(128 -> 512, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (7): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=512) ) (conv1): Conv3D(512 -> 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (conv2): Conv3D(128 -> 128, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1), bias=False) (bn1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=128) (bn2): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=128) (conv3): Conv3D(128 -> 512, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (bn3): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=512) (relu): Activation(relu) ) ) (2): HybridSequential( (0): Bottleneck( (bottleneck): HybridSequential( (0): Conv3D(512 -> 256, kernel_size=(3, 1, 1), stride=(1, 1, 1), padding=(1, 0, 0), bias=False) (1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=256) (2): Activation(relu) (3): Conv3D(256 -> 256, kernel_size=(1, 3, 3), stride=(1, 2, 2), padding=(0, 1, 1), bias=False) (4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=256) (5): Activation(relu) (6): Conv3D(256 -> 1024, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (7): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=1024) ) (conv1): Conv3D(512 -> 256, kernel_size=(3, 1, 1), stride=(1, 1, 1), padding=(1, 0, 0), bias=False) (conv2): Conv3D(256 -> 256, kernel_size=(1, 3, 3), stride=(1, 2, 2), padding=(0, 1, 1), bias=False) (bn1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=256) (bn2): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=256) (conv3): Conv3D(256 -> 1024, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (bn3): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=1024) (relu): Activation(relu) (downsample): HybridSequential( (0): Conv3D(512 -> 1024, kernel_size=(1, 1, 1), stride=(1, 2, 2), bias=False) (1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=1024) ) ) (1): Bottleneck( (bottleneck): HybridSequential( (0): Conv3D(1024 -> 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=256) (2): Activation(relu) (3): Conv3D(256 -> 256, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1), bias=False) (4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=256) (5): Activation(relu) (6): Conv3D(256 -> 1024, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (7): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=1024) ) (conv1): Conv3D(1024 -> 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (conv2): Conv3D(256 -> 256, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1), bias=False) (bn1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=256) (bn2): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=256) (conv3): Conv3D(256 -> 1024, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (bn3): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=1024) (relu): Activation(relu) ) (2): Bottleneck( (bottleneck): HybridSequential( (0): Conv3D(1024 -> 256, kernel_size=(3, 1, 1), stride=(1, 1, 1), padding=(1, 0, 0), bias=False) (1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=256) (2): Activation(relu) (3): Conv3D(256 -> 256, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1), bias=False) (4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=256) (5): Activation(relu) (6): Conv3D(256 -> 1024, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (7): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=1024) ) (conv1): Conv3D(1024 -> 256, kernel_size=(3, 1, 1), stride=(1, 1, 1), padding=(1, 0, 0), bias=False) (conv2): Conv3D(256 -> 256, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1), bias=False) (bn1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=256) (bn2): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=256) (conv3): Conv3D(256 -> 1024, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (bn3): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=1024) (relu): Activation(relu) ) (3): Bottleneck( (bottleneck): HybridSequential( (0): Conv3D(1024 -> 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=256) (2): Activation(relu) (3): Conv3D(256 -> 256, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1), bias=False) (4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=256) (5): Activation(relu) (6): Conv3D(256 -> 1024, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (7): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=1024) ) (conv1): Conv3D(1024 -> 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (conv2): Conv3D(256 -> 256, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1), bias=False) (bn1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=256) (bn2): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=256) (conv3): Conv3D(256 -> 1024, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (bn3): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=1024) (relu): Activation(relu) ) (4): Bottleneck( (bottleneck): HybridSequential( (0): Conv3D(1024 -> 256, kernel_size=(3, 1, 1), stride=(1, 1, 1), padding=(1, 0, 0), bias=False) (1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=256) (2): Activation(relu) (3): Conv3D(256 -> 256, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1), bias=False) (4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=256) (5): Activation(relu) (6): Conv3D(256 -> 1024, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (7): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=1024) ) (conv1): Conv3D(1024 -> 256, kernel_size=(3, 1, 1), stride=(1, 1, 1), padding=(1, 0, 0), bias=False) (conv2): Conv3D(256 -> 256, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1), bias=False) (bn1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=256) (bn2): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=256) (conv3): Conv3D(256 -> 1024, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (bn3): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=1024) (relu): Activation(relu) ) (5): Bottleneck( (bottleneck): HybridSequential( (0): Conv3D(1024 -> 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=256) (2): Activation(relu) (3): Conv3D(256 -> 256, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1), bias=False) (4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=256) (5): Activation(relu) (6): Conv3D(256 -> 1024, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (7): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=1024) ) (conv1): Conv3D(1024 -> 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (conv2): Conv3D(256 -> 256, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1), bias=False) (bn1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=256) (bn2): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=256) (conv3): Conv3D(256 -> 1024, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (bn3): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=1024) (relu): Activation(relu) ) ) (3): HybridSequential( (0): Bottleneck( (bottleneck): HybridSequential( (0): Conv3D(1024 -> 512, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=512) (2): Activation(relu) (3): Conv3D(512 -> 512, kernel_size=(1, 3, 3), stride=(1, 2, 2), padding=(0, 1, 1), bias=False) (4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=512) (5): Activation(relu) (6): Conv3D(512 -> 2048, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (7): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=2048) ) (conv1): Conv3D(1024 -> 512, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (conv2): Conv3D(512 -> 512, kernel_size=(1, 3, 3), stride=(1, 2, 2), padding=(0, 1, 1), bias=False) (bn1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=512) (bn2): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=512) (conv3): Conv3D(512 -> 2048, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (bn3): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=2048) (relu): Activation(relu) (downsample): HybridSequential( (0): Conv3D(1024 -> 2048, kernel_size=(1, 1, 1), stride=(1, 2, 2), bias=False) (1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=2048) ) ) (1): Bottleneck( (bottleneck): HybridSequential( (0): Conv3D(2048 -> 512, kernel_size=(3, 1, 1), stride=(1, 1, 1), padding=(1, 0, 0), bias=False) (1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=512) (2): Activation(relu) (3): Conv3D(512 -> 512, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1), bias=False) (4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=512) (5): Activation(relu) (6): Conv3D(512 -> 2048, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (7): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=2048) ) (conv1): Conv3D(2048 -> 512, kernel_size=(3, 1, 1), stride=(1, 1, 1), padding=(1, 0, 0), bias=False) (conv2): Conv3D(512 -> 512, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1), bias=False) (bn1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=512) (bn2): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=512) (conv3): Conv3D(512 -> 2048, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (bn3): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=2048) (relu): Activation(relu) ) (2): Bottleneck( (bottleneck): HybridSequential( (0): Conv3D(2048 -> 512, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=512) (2): Activation(relu) (3): Conv3D(512 -> 512, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1), bias=False) (4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=512) (5): Activation(relu) (6): Conv3D(512 -> 2048, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (7): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=2048) ) (conv1): Conv3D(2048 -> 512, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (conv2): Conv3D(512 -> 512, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1), bias=False) (bn1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=512) (bn2): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=512) (conv3): Conv3D(512 -> 2048, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False) (bn3): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=2048) (relu): Activation(relu) ) ) ) (st_avg): 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 = 0.8, axes=()) (1): Dense(2048 -> 101, linear) ) (fc): Dense(2048 -> 101, linear) ) Load 9537 training samples and 3783 validation samples. Epoch[000] Batch [0019]/[0149] Speed: 26.726844 samples/sec accuracy=2.265625 loss=4.600650 lr=0.001000 Epoch[000] Batch [0039]/[0149] Speed: 114.654427 samples/sec accuracy=3.203125 loss=4.579816 lr=0.001000 Epoch[000] Batch [0059]/[0149] Speed: 106.924243 samples/sec accuracy=5.130208 loss=4.549021 lr=0.001000 Epoch[000] Batch [0079]/[0149] Speed: 108.818531 samples/sec accuracy=7.109375 loss=4.516935 lr=0.001000 Epoch[000] Batch [0099]/[0149] Speed: 110.825769 samples/sec accuracy=8.500000 loss=4.485839 lr=0.001000 Epoch[000] Batch [0119]/[0149] Speed: 110.340774 samples/sec accuracy=9.778646 loss=4.450389 lr=0.001000 Epoch[000] Batch [0139]/[0149] Speed: 129.216043 samples/sec accuracy=11.238839 loss=4.415392 lr=0.001000 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 000] training: accuracy=11.902265 loss=4.398111 [Epoch 000] speed: 78 samples/sec time cost: 165.429550 [Epoch 000] validation: acc-top1=28.972458 acc-top5=56.647246 loss=4.019325 Epoch[001] Batch [0019]/[0149] Speed: 49.723925 samples/sec accuracy=24.531250 loss=4.026907 lr=0.001000 Epoch[001] Batch [0039]/[0149] Speed: 113.867123 samples/sec accuracy=25.234375 loss=3.967111 lr=0.001000 Epoch[001] Batch [0059]/[0149] Speed: 108.980297 samples/sec accuracy=25.442708 loss=3.911245 lr=0.001000 Epoch[001] Batch [0079]/[0149] Speed: 114.308670 samples/sec accuracy=25.234375 loss=3.863987 lr=0.001000 Epoch[001] Batch [0099]/[0149] Speed: 107.816904 samples/sec accuracy=25.578125 loss=3.808101 lr=0.001000 Epoch[001] Batch [0119]/[0149] Speed: 114.666820 samples/sec accuracy=26.236979 loss=3.749160 lr=0.001000 Epoch[001] Batch [0139]/[0149] Speed: 128.584450 samples/sec accuracy=26.941964 loss=3.690502 lr=0.001000 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 001] training: accuracy=27.307047 loss=3.666058 [Epoch 001] speed: 97 samples/sec time cost: 133.374785 [Epoch 001] validation: acc-top1=41.498941 acc-top5=72.086864 loss=3.015836 Epoch[002] Batch [0019]/[0149] Speed: 51.076685 samples/sec accuracy=34.609375 loss=3.155439 lr=0.001000 Epoch[002] Batch [0039]/[0149] Speed: 113.381080 samples/sec accuracy=34.921875 loss=3.111871 lr=0.001000 Epoch[002] Batch [0059]/[0149] Speed: 109.782009 samples/sec accuracy=35.755208 loss=3.060181 lr=0.001000 Epoch[002] Batch [0079]/[0149] Speed: 110.074468 samples/sec accuracy=36.582031 loss=3.003101 lr=0.001000 Epoch[002] Batch [0099]/[0149] Speed: 113.285305 samples/sec accuracy=36.968750 loss=2.965377 lr=0.001000 Epoch[002] Batch [0119]/[0149] Speed: 112.176956 samples/sec accuracy=37.486979 loss=2.923254 lr=0.001000 Epoch[002] Batch [0139]/[0149] Speed: 128.443576 samples/sec accuracy=38.046875 loss=2.888220 lr=0.001000 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 002] training: accuracy=38.380872 loss=2.870187 [Epoch 002] speed: 97 samples/sec time cost: 131.323919 [Epoch 002] validation: acc-top1=51.933263 acc-top5=80.323093 loss=2.221529 Epoch[003] Batch [0019]/[0149] Speed: 50.814401 samples/sec accuracy=44.453125 loss=2.526984 lr=0.001000 Epoch[003] Batch [0039]/[0149] Speed: 115.732332 samples/sec accuracy=44.804688 loss=2.488705 lr=0.001000 Epoch[003] Batch [0059]/[0149] Speed: 109.824449 samples/sec accuracy=46.041667 loss=2.435979 lr=0.001000 Epoch[003] Batch [0079]/[0149] Speed: 111.735549 samples/sec accuracy=46.035156 loss=2.406014 lr=0.001000 Epoch[003] Batch [0099]/[0149] Speed: 109.179674 samples/sec accuracy=46.625000 loss=2.374368 lr=0.001000 Epoch[003] Batch [0119]/[0149] Speed: 113.559412 samples/sec accuracy=46.848958 loss=2.343472 lr=0.001000 Epoch[003] Batch [0139]/[0149] Speed: 127.368194 samples/sec accuracy=47.243304 loss=2.307256 lr=0.001000 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 003] training: accuracy=47.462248 loss=2.297749 [Epoch 003] speed: 97 samples/sec time cost: 132.666934 [Epoch 003] validation: acc-top1=58.739407 acc-top5=85.858051 loss=1.698563 Epoch[004] Batch [0019]/[0149] Speed: 51.118901 samples/sec accuracy=52.968750 loss=2.061869 lr=0.001000 Epoch[004] Batch [0039]/[0149] Speed: 112.993691 samples/sec accuracy=53.125000 loss=2.016305 lr=0.001000 Epoch[004] Batch [0059]/[0149] Speed: 108.217153 samples/sec accuracy=53.281250 loss=2.002605 lr=0.001000 Epoch[004] Batch [0079]/[0149] Speed: 110.761872 samples/sec accuracy=54.492188 loss=1.964291 lr=0.001000 Epoch[004] Batch [0099]/[0149] Speed: 108.968363 samples/sec accuracy=54.734375 loss=1.943430 lr=0.001000 Epoch[004] Batch [0119]/[0149] Speed: 111.143361 samples/sec accuracy=55.377604 loss=1.914870 lr=0.001000 Epoch[004] Batch [0139]/[0149] Speed: 124.896810 samples/sec accuracy=55.368304 loss=1.905468 lr=0.001000 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 004] training: accuracy=55.536913 loss=1.893636 [Epoch 004] speed: 96 samples/sec time cost: 132.838347 [Epoch 004] validation: acc-top1=61.493644 acc-top5=86.943856 loss=1.543145 Epoch[005] Batch [0019]/[0149] Speed: 52.402166 samples/sec accuracy=57.890625 loss=1.715846 lr=0.001000 Epoch[005] Batch [0039]/[0149] Speed: 114.478440 samples/sec accuracy=59.140625 loss=1.695047 lr=0.001000 Epoch[005] Batch [0059]/[0149] Speed: 111.032772 samples/sec accuracy=59.140625 loss=1.683172 lr=0.001000 Epoch[005] Batch [0079]/[0149] Speed: 111.970820 samples/sec accuracy=59.257812 loss=1.668745 lr=0.001000 Epoch[005] Batch [0099]/[0149] Speed: 113.097776 samples/sec accuracy=59.656250 loss=1.645224 lr=0.001000 Epoch[005] Batch [0119]/[0149] Speed: 110.828521 samples/sec accuracy=60.013021 loss=1.630153 lr=0.001000 Epoch[005] Batch [0139]/[0149] Speed: 127.041806 samples/sec accuracy=60.223214 loss=1.616554 lr=0.001000 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 005] training: accuracy=60.539010 loss=1.608307 [Epoch 005] speed: 98 samples/sec time cost: 132.905426 [Epoch 005] validation: acc-top1=66.975636 acc-top5=90.519068 loss=1.210191 Epoch[006] Batch [0019]/[0149] Speed: 52.990093 samples/sec accuracy=66.718750 loss=1.379114 lr=0.001000 Epoch[006] Batch [0039]/[0149] Speed: 114.954865 samples/sec accuracy=65.781250 loss=1.413051 lr=0.001000 Epoch[006] Batch [0059]/[0149] Speed: 111.407240 samples/sec accuracy=66.015625 loss=1.402617 lr=0.001000 Epoch[006] Batch [0079]/[0149] Speed: 114.751523 samples/sec accuracy=65.957031 loss=1.401639 lr=0.001000 Epoch[006] Batch [0099]/[0149] Speed: 108.650953 samples/sec accuracy=65.703125 loss=1.394485 lr=0.001000 Epoch[006] Batch [0119]/[0149] Speed: 113.602039 samples/sec accuracy=65.807292 loss=1.386242 lr=0.001000 Epoch[006] Batch [0139]/[0149] Speed: 121.675789 samples/sec accuracy=65.881696 loss=1.375350 lr=0.001000 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 006] training: accuracy=65.782299 loss=1.374916 [Epoch 006] speed: 98 samples/sec time cost: 130.588896 [Epoch 006] validation: acc-top1=68.405720 acc-top5=91.551907 loss=1.129389 Epoch[007] Batch [0019]/[0149] Speed: 50.092781 samples/sec accuracy=66.796875 loss=1.307842 lr=0.001000 Epoch[007] Batch [0039]/[0149] Speed: 114.127965 samples/sec accuracy=66.523438 loss=1.301062 lr=0.001000 Epoch[007] Batch [0059]/[0149] Speed: 111.042013 samples/sec accuracy=67.317708 loss=1.279591 lr=0.001000 Epoch[007] Batch [0079]/[0149] Speed: 113.687225 samples/sec accuracy=67.207031 loss=1.269803 lr=0.001000 Epoch[007] Batch [0099]/[0149] Speed: 111.431865 samples/sec accuracy=67.328125 loss=1.268623 lr=0.001000 Epoch[007] Batch [0119]/[0149] Speed: 109.706895 samples/sec accuracy=67.916667 loss=1.247737 lr=0.001000 Epoch[007] Batch [0139]/[0149] Speed: 125.241030 samples/sec accuracy=68.080357 loss=1.239855 lr=0.001000 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 007] training: accuracy=68.320050 loss=1.238772 [Epoch 007] speed: 97 samples/sec time cost: 132.267603 [Epoch 007] validation: acc-top1=71.451271 acc-top5=92.293432 loss=1.031536 Epoch[008] Batch [0019]/[0149] Speed: 52.028056 samples/sec accuracy=72.500000 loss=1.119752 lr=0.001000 Epoch[008] Batch [0039]/[0149] Speed: 116.197845 samples/sec accuracy=70.898438 loss=1.128009 lr=0.001000 Epoch[008] Batch [0059]/[0149] Speed: 109.303995 samples/sec accuracy=71.822917 loss=1.102774 lr=0.001000 Epoch[008] Batch [0079]/[0149] Speed: 114.476877 samples/sec accuracy=71.972656 loss=1.098524 lr=0.001000 Epoch[008] Batch [0099]/[0149] Speed: 104.832764 samples/sec accuracy=71.968750 loss=1.096473 lr=0.001000 Epoch[008] Batch [0119]/[0149] Speed: 115.882168 samples/sec accuracy=71.914062 loss=1.090866 lr=0.001000 Epoch[008] Batch [0139]/[0149] Speed: 125.494144 samples/sec accuracy=71.808036 loss=1.092275 lr=0.001000 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 008] training: accuracy=71.906460 loss=1.087141 [Epoch 008] speed: 98 samples/sec time cost: 131.608266 [Epoch 008] validation: acc-top1=71.345339 acc-top5=92.346398 loss=1.022967 Epoch[009] Batch [0019]/[0149] Speed: 51.640313 samples/sec accuracy=73.671875 loss=1.018197 lr=0.001000 Epoch[009] Batch [0039]/[0149] Speed: 117.098933 samples/sec accuracy=74.140625 loss=1.019635 lr=0.001000 Epoch[009] Batch [0059]/[0149] Speed: 109.090991 samples/sec accuracy=73.723958 loss=1.020854 lr=0.001000 Epoch[009] Batch [0079]/[0149] Speed: 110.664586 samples/sec accuracy=74.433594 loss=1.001901 lr=0.001000 Epoch[009] Batch [0099]/[0149] Speed: 109.111550 samples/sec accuracy=74.234375 loss=1.005463 lr=0.001000 Epoch[009] Batch [0119]/[0149] Speed: 111.278350 samples/sec accuracy=74.518229 loss=0.996760 lr=0.001000 Epoch[009] Batch [0139]/[0149] Speed: 126.724945 samples/sec accuracy=74.854911 loss=0.989645 lr=0.001000 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 009] training: accuracy=74.695889 loss=0.986131 [Epoch 009] speed: 97 samples/sec time cost: 131.886641 [Epoch 009] validation: acc-top1=73.411017 acc-top5=93.061441 loss=0.943583 Epoch[010] Batch [0019]/[0149] Speed: 51.537894 samples/sec accuracy=75.546875 loss=0.889018 lr=0.001000 Epoch[010] Batch [0039]/[0149] Speed: 117.967745 samples/sec accuracy=75.351562 loss=0.908673 lr=0.001000 Epoch[010] Batch [0059]/[0149] Speed: 111.373403 samples/sec accuracy=75.104167 loss=0.918537 lr=0.001000 Epoch[010] Batch [0079]/[0149] Speed: 114.150331 samples/sec accuracy=75.917969 loss=0.896578 lr=0.001000 Epoch[010] Batch [0099]/[0149] Speed: 107.025246 samples/sec accuracy=76.593750 loss=0.885226 lr=0.001000 Epoch[010] Batch [0119]/[0149] Speed: 114.694769 samples/sec accuracy=76.588542 loss=0.883174 lr=0.001000 Epoch[010] Batch [0139]/[0149] Speed: 129.794130 samples/sec accuracy=76.729911 loss=0.879500 lr=0.001000 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 010] training: accuracy=76.845638 loss=0.875107 [Epoch 010] speed: 98 samples/sec time cost: 131.668121 [Epoch 010] validation: acc-top1=73.940678 acc-top5=93.114407 loss=0.910229 Epoch[011] Batch [0019]/[0149] Speed: 51.268142 samples/sec accuracy=79.062500 loss=0.811543 lr=0.001000 Epoch[011] Batch [0039]/[0149] Speed: 114.588855 samples/sec accuracy=78.437500 loss=0.832904 lr=0.001000 Epoch[011] Batch [0059]/[0149] Speed: 110.809848 samples/sec accuracy=78.411458 loss=0.826264 lr=0.001000 Epoch[011] Batch [0079]/[0149] Speed: 114.137845 samples/sec accuracy=78.339844 loss=0.824435 lr=0.001000 Epoch[011] Batch [0099]/[0149] Speed: 106.855735 samples/sec accuracy=78.359375 loss=0.819642 lr=0.001000 Epoch[011] Batch [0119]/[0149] Speed: 117.387814 samples/sec accuracy=78.098958 loss=0.819696 lr=0.001000 Epoch[011] Batch [0139]/[0149] Speed: 126.987519 samples/sec accuracy=78.437500 loss=0.809987 lr=0.001000 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 011] training: accuracy=78.450084 loss=0.810003 [Epoch 011] speed: 98 samples/sec time cost: 131.019871 [Epoch 011] validation: acc-top1=73.940678 acc-top5=92.664195 loss=0.937576 Epoch[012] Batch [0019]/[0149] Speed: 50.941351 samples/sec accuracy=79.375000 loss=0.771381 lr=0.001000 Epoch[012] Batch [0039]/[0149] Speed: 115.535817 samples/sec accuracy=80.195312 loss=0.769579 lr=0.001000 Epoch[012] Batch [0059]/[0149] Speed: 108.199919 samples/sec accuracy=80.572917 loss=0.744197 lr=0.001000 Epoch[012] Batch [0079]/[0149] Speed: 109.219041 samples/sec accuracy=80.449219 loss=0.741647 lr=0.001000 Epoch[012] Batch [0099]/[0149] Speed: 111.424191 samples/sec accuracy=80.828125 loss=0.734540 lr=0.001000 Epoch[012] Batch [0119]/[0149] Speed: 111.898925 samples/sec accuracy=80.716146 loss=0.739561 lr=0.001000 Epoch[012] Batch [0139]/[0149] Speed: 131.517210 samples/sec accuracy=80.736607 loss=0.736452 lr=0.001000 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 012] training: accuracy=80.861997 loss=0.733505 [Epoch 012] speed: 97 samples/sec time cost: 131.878455 [Epoch 012] validation: acc-top1=74.496822 acc-top5=92.929025 loss=0.944700 Epoch[013] Batch [0019]/[0149] Speed: 51.685199 samples/sec accuracy=82.031250 loss=0.700854 lr=0.001000 Epoch[013] Batch [0039]/[0149] Speed: 115.847575 samples/sec accuracy=82.187500 loss=0.688300 lr=0.001000 Epoch[013] Batch [0059]/[0149] Speed: 112.051087 samples/sec accuracy=81.796875 loss=0.698649 lr=0.001000 Epoch[013] Batch [0079]/[0149] Speed: 111.965918 samples/sec accuracy=81.855469 loss=0.691987 lr=0.001000 Epoch[013] Batch [0099]/[0149] Speed: 110.263512 samples/sec accuracy=81.750000 loss=0.696812 lr=0.001000 Epoch[013] Batch [0119]/[0149] Speed: 114.248285 samples/sec accuracy=81.822917 loss=0.694345 lr=0.001000 Epoch[013] Batch [0139]/[0149] Speed: 129.275973 samples/sec accuracy=81.964286 loss=0.689995 lr=0.001000 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 013] training: accuracy=81.795302 loss=0.692379 [Epoch 013] speed: 98 samples/sec time cost: 131.229178 [Epoch 013] validation: acc-top1=76.721398 acc-top5=93.697034 loss=0.895697 Epoch[014] Batch [0019]/[0149] Speed: 52.168709 samples/sec accuracy=83.750000 loss=0.637656 lr=0.001000 Epoch[014] Batch [0039]/[0149] Speed: 116.274160 samples/sec accuracy=83.710938 loss=0.633278 lr=0.001000 Epoch[014] Batch [0059]/[0149] Speed: 110.716885 samples/sec accuracy=83.723958 loss=0.635128 lr=0.001000 Epoch[014] Batch [0079]/[0149] Speed: 109.849579 samples/sec accuracy=83.535156 loss=0.633277 lr=0.001000 Epoch[014] Batch [0099]/[0149] Speed: 112.521800 samples/sec accuracy=83.750000 loss=0.631006 lr=0.001000 Epoch[014] Batch [0119]/[0149] Speed: 112.961086 samples/sec accuracy=83.828125 loss=0.629172 lr=0.001000 Epoch[014] Batch [0139]/[0149] Speed: 128.839380 samples/sec accuracy=83.761161 loss=0.631352 lr=0.001000 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 014] training: accuracy=83.724832 loss=0.631288 [Epoch 014] speed: 98 samples/sec time cost: 131.840890 [Epoch 014] validation: acc-top1=75.582627 acc-top5=93.776483 loss=0.881868 Epoch[015] Batch [0019]/[0149] Speed: 51.079684 samples/sec accuracy=83.281250 loss=0.617423 lr=0.001000 Epoch[015] Batch [0039]/[0149] Speed: 114.879305 samples/sec accuracy=84.296875 loss=0.596530 lr=0.001000 Epoch[015] Batch [0059]/[0149] Speed: 110.231817 samples/sec accuracy=84.661458 loss=0.584831 lr=0.001000 Epoch[015] Batch [0079]/[0149] Speed: 111.320769 samples/sec accuracy=84.550781 loss=0.580729 lr=0.001000 Epoch[015] Batch [0099]/[0149] Speed: 110.146237 samples/sec accuracy=84.843750 loss=0.573462 lr=0.001000 Epoch[015] Batch [0119]/[0149] Speed: 113.222930 samples/sec accuracy=84.804688 loss=0.576835 lr=0.001000 Epoch[015] Batch [0139]/[0149] Speed: 128.943531 samples/sec accuracy=84.899554 loss=0.571302 lr=0.001000 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 015] training: accuracy=85.088087 loss=0.567874 [Epoch 015] speed: 97 samples/sec time cost: 132.599587 [Epoch 015] validation: acc-top1=77.224576 acc-top5=94.226695 loss=0.826481 Epoch[016] Batch [0019]/[0149] Speed: 53.064089 samples/sec accuracy=86.171875 loss=0.529492 lr=0.001000 Epoch[016] Batch [0039]/[0149] Speed: 116.520296 samples/sec accuracy=87.109375 loss=0.506874 lr=0.001000 Epoch[016] Batch [0059]/[0149] Speed: 109.767305 samples/sec accuracy=86.744792 loss=0.526339 lr=0.001000 Epoch[016] Batch [0079]/[0149] Speed: 113.177733 samples/sec accuracy=86.503906 loss=0.529881 lr=0.001000 Epoch[016] Batch [0099]/[0149] Speed: 111.288587 samples/sec accuracy=86.375000 loss=0.527921 lr=0.001000 Epoch[016] Batch [0119]/[0149] Speed: 111.802659 samples/sec accuracy=86.093750 loss=0.536082 lr=0.001000 Epoch[016] Batch [0139]/[0149] Speed: 129.901114 samples/sec accuracy=86.082589 loss=0.539385 lr=0.001000 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 016] training: accuracy=86.168205 loss=0.535638 [Epoch 016] speed: 99 samples/sec time cost: 131.103240 [Epoch 016] validation: acc-top1=77.383475 acc-top5=94.412076 loss=0.841957 Epoch[017] Batch [0019]/[0149] Speed: 52.652230 samples/sec accuracy=87.031250 loss=0.506507 lr=0.001000 Epoch[017] Batch [0039]/[0149] Speed: 117.215236 samples/sec accuracy=86.445312 loss=0.510793 lr=0.001000 Epoch[017] Batch [0059]/[0149] Speed: 111.402508 samples/sec accuracy=86.328125 loss=0.513151 lr=0.001000 Epoch[017] Batch [0079]/[0149] Speed: 114.783820 samples/sec accuracy=86.230469 loss=0.510653 lr=0.001000 Epoch[017] Batch [0099]/[0149] Speed: 108.740256 samples/sec accuracy=86.265625 loss=0.511785 lr=0.001000 Epoch[017] Batch [0119]/[0149] Speed: 116.665307 samples/sec accuracy=86.328125 loss=0.508357 lr=0.001000 Epoch[017] Batch [0139]/[0149] Speed: 127.881479 samples/sec accuracy=86.506696 loss=0.506337 lr=0.001000 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 017] training: accuracy=86.755453 loss=0.500712 [Epoch 017] speed: 99 samples/sec time cost: 129.476349 [Epoch 017] validation: acc-top1=76.747881 acc-top5=93.802966 loss=0.843373 Epoch[018] Batch [0019]/[0149] Speed: 54.250085 samples/sec accuracy=87.343750 loss=0.456343 lr=0.001000 Epoch[018] Batch [0039]/[0149] Speed: 115.165042 samples/sec accuracy=88.203125 loss=0.451884 lr=0.001000 Epoch[018] Batch [0059]/[0149] Speed: 111.091026 samples/sec accuracy=88.072917 loss=0.464805 lr=0.001000 Epoch[018] Batch [0079]/[0149] Speed: 113.309719 samples/sec accuracy=87.617188 loss=0.467597 lr=0.001000 Epoch[018] Batch [0099]/[0149] Speed: 110.182903 samples/sec accuracy=88.000000 loss=0.458726 lr=0.001000 Epoch[018] Batch [0119]/[0149] Speed: 110.178998 samples/sec accuracy=87.916667 loss=0.459373 lr=0.001000 Epoch[018] Batch [0139]/[0149] Speed: 127.937616 samples/sec accuracy=87.845982 loss=0.461523 lr=0.001000 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 018] training: accuracy=87.772651 loss=0.463164 [Epoch 018] speed: 99 samples/sec time cost: 131.147446 [Epoch 018] validation: acc-top1=78.522246 acc-top5=94.650424 loss=0.833988 Epoch[019] Batch [0019]/[0149] Speed: 53.242053 samples/sec accuracy=89.062500 loss=0.432519 lr=0.001000 Epoch[019] Batch [0039]/[0149] Speed: 114.736620 samples/sec accuracy=88.984375 loss=0.432257 lr=0.001000 Epoch[019] Batch [0059]/[0149] Speed: 114.748906 samples/sec accuracy=89.036458 loss=0.441426 lr=0.001000 Epoch[019] Batch [0079]/[0149] Speed: 112.056429 samples/sec accuracy=89.062500 loss=0.440027 lr=0.001000 Epoch[019] Batch [0099]/[0149] Speed: 110.571426 samples/sec accuracy=88.890625 loss=0.437897 lr=0.001000 Epoch[019] Batch [0119]/[0149] Speed: 111.000199 samples/sec accuracy=88.854167 loss=0.436175 lr=0.001000 Epoch[019] Batch [0139]/[0149] Speed: 130.413666 samples/sec accuracy=88.861607 loss=0.437309 lr=0.001000 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 019] training: accuracy=88.873742 loss=0.436168 [Epoch 019] speed: 99 samples/sec time cost: 130.331631 [Epoch 019] validation: acc-top1=78.575212 acc-top5=94.941737 loss=0.830529 Epoch[020] Batch [0019]/[0149] Speed: 51.366753 samples/sec accuracy=89.687500 loss=0.401510 lr=0.000100 Epoch[020] Batch [0039]/[0149] Speed: 115.323830 samples/sec accuracy=89.453125 loss=0.415057 lr=0.000100 Epoch[020] Batch [0059]/[0149] Speed: 108.029652 samples/sec accuracy=89.270833 loss=0.415006 lr=0.000100 Epoch[020] Batch [0079]/[0149] Speed: 114.551637 samples/sec accuracy=89.980469 loss=0.403228 lr=0.000100 Epoch[020] Batch [0099]/[0149] Speed: 108.869136 samples/sec accuracy=90.281250 loss=0.392647 lr=0.000100 Epoch[020] Batch [0119]/[0149] Speed: 113.229648 samples/sec accuracy=90.104167 loss=0.395237 lr=0.000100 Epoch[020] Batch [0139]/[0149] Speed: 127.772375 samples/sec accuracy=90.011161 loss=0.399340 lr=0.000100 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 020] training: accuracy=89.985319 loss=0.398132 [Epoch 020] speed: 97 samples/sec time cost: 132.729985 [Epoch 020] validation: acc-top1=79.449153 acc-top5=94.835805 loss=0.774532 Epoch[021] Batch [0019]/[0149] Speed: 52.124259 samples/sec accuracy=90.312500 loss=0.357000 lr=0.000100 Epoch[021] Batch [0039]/[0149] Speed: 114.800839 samples/sec accuracy=90.312500 loss=0.381794 lr=0.000100 Epoch[021] Batch [0059]/[0149] Speed: 109.355365 samples/sec accuracy=90.026042 loss=0.387623 lr=0.000100 Epoch[021] Batch [0079]/[0149] Speed: 112.166789 samples/sec accuracy=90.234375 loss=0.382472 lr=0.000100 Epoch[021] Batch [0099]/[0149] Speed: 108.108346 samples/sec accuracy=90.421875 loss=0.379072 lr=0.000100 Epoch[021] Batch [0119]/[0149] Speed: 110.380876 samples/sec accuracy=90.234375 loss=0.387843 lr=0.000100 Epoch[021] Batch [0139]/[0149] Speed: 125.322321 samples/sec accuracy=90.390625 loss=0.383730 lr=0.000100 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 021] training: accuracy=90.341862 loss=0.383732 [Epoch 021] speed: 97 samples/sec time cost: 132.008792 [Epoch 021] validation: acc-top1=79.661017 acc-top5=95.180085 loss=0.769362 Epoch[022] Batch [0019]/[0149] Speed: 54.319772 samples/sec accuracy=90.390625 loss=0.382273 lr=0.000100 Epoch[022] Batch [0039]/[0149] Speed: 111.872663 samples/sec accuracy=90.039062 loss=0.382745 lr=0.000100 Epoch[022] Batch [0059]/[0149] Speed: 112.012284 samples/sec accuracy=89.817708 loss=0.382337 lr=0.000100 Epoch[022] Batch [0079]/[0149] Speed: 114.197239 samples/sec accuracy=89.980469 loss=0.383126 lr=0.000100 Epoch[022] Batch [0099]/[0149] Speed: 106.758186 samples/sec accuracy=89.765625 loss=0.389610 lr=0.000100 Epoch[022] Batch [0119]/[0149] Speed: 114.859655 samples/sec accuracy=89.726562 loss=0.389814 lr=0.000100 Epoch[022] Batch [0139]/[0149] Speed: 128.609610 samples/sec accuracy=89.854911 loss=0.385066 lr=0.000100 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 022] training: accuracy=89.796560 loss=0.383906 [Epoch 022] speed: 99 samples/sec time cost: 131.978575 [Epoch 022] validation: acc-top1=79.316737 acc-top5=94.915254 loss=0.772482 Epoch[023] Batch [0019]/[0149] Speed: 52.366004 samples/sec accuracy=89.765625 loss=0.394372 lr=0.000100 Epoch[023] Batch [0039]/[0149] Speed: 114.606805 samples/sec accuracy=90.117188 loss=0.390251 lr=0.000100 Epoch[023] Batch [0059]/[0149] Speed: 110.559968 samples/sec accuracy=89.947917 loss=0.393180 lr=0.000100 Epoch[023] Batch [0079]/[0149] Speed: 111.331263 samples/sec accuracy=89.804688 loss=0.394974 lr=0.000100 Epoch[023] Batch [0099]/[0149] Speed: 112.827992 samples/sec accuracy=89.984375 loss=0.391501 lr=0.000100 Epoch[023] Batch [0119]/[0149] Speed: 109.948195 samples/sec accuracy=89.843750 loss=0.397223 lr=0.000100 Epoch[023] Batch [0139]/[0149] Speed: 126.952884 samples/sec accuracy=89.866071 loss=0.393762 lr=0.000100 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 023] training: accuracy=89.953859 loss=0.393856 [Epoch 023] speed: 98 samples/sec time cost: 132.049267 [Epoch 023] validation: acc-top1=79.687500 acc-top5=95.444915 loss=0.758913 Epoch[024] Batch [0019]/[0149] Speed: 51.461595 samples/sec accuracy=89.687500 loss=0.378520 lr=0.000100 Epoch[024] Batch [0039]/[0149] Speed: 118.125950 samples/sec accuracy=89.140625 loss=0.399831 lr=0.000100 Epoch[024] Batch [0059]/[0149] Speed: 110.043997 samples/sec accuracy=90.104167 loss=0.384489 lr=0.000100 Epoch[024] Batch [0079]/[0149] Speed: 113.086381 samples/sec accuracy=90.195312 loss=0.377037 lr=0.000100 Epoch[024] Batch [0099]/[0149] Speed: 109.219772 samples/sec accuracy=90.234375 loss=0.375328 lr=0.000100 Epoch[024] Batch [0119]/[0149] Speed: 111.953354 samples/sec accuracy=90.091146 loss=0.378697 lr=0.000100 Epoch[024] Batch [0139]/[0149] Speed: 126.861760 samples/sec accuracy=90.167411 loss=0.374543 lr=0.000100 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 024] training: accuracy=90.184564 loss=0.375775 [Epoch 024] speed: 97 samples/sec time cost: 132.260123 [Epoch 024] validation: acc-top1=79.475636 acc-top5=95.338983 loss=0.765256 Epoch[025] Batch [0019]/[0149] Speed: 51.804462 samples/sec accuracy=91.093750 loss=0.353217 lr=0.000100 Epoch[025] Batch [0039]/[0149] Speed: 114.824246 samples/sec accuracy=91.015625 loss=0.363959 lr=0.000100 Epoch[025] Batch [0059]/[0149] Speed: 111.114800 samples/sec accuracy=90.937500 loss=0.366825 lr=0.000100 Epoch[025] Batch [0079]/[0149] Speed: 110.688283 samples/sec accuracy=91.074219 loss=0.356930 lr=0.000100 Epoch[025] Batch [0099]/[0149] Speed: 110.851489 samples/sec accuracy=90.906250 loss=0.357990 lr=0.000100 Epoch[025] Batch [0119]/[0149] Speed: 113.087339 samples/sec accuracy=90.976562 loss=0.362540 lr=0.000100 Epoch[025] Batch [0139]/[0149] Speed: 128.625577 samples/sec accuracy=90.937500 loss=0.364764 lr=0.000100 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 025] training: accuracy=90.918624 loss=0.364756 [Epoch 025] speed: 98 samples/sec time cost: 133.867804 [Epoch 025] validation: acc-top1=79.687500 acc-top5=95.153602 loss=0.752365 Epoch[026] Batch [0019]/[0149] Speed: 51.902868 samples/sec accuracy=91.171875 loss=0.350554 lr=0.000100 Epoch[026] Batch [0039]/[0149] Speed: 118.258186 samples/sec accuracy=90.898438 loss=0.352244 lr=0.000100 Epoch[026] Batch [0059]/[0149] Speed: 110.366905 samples/sec accuracy=90.338542 loss=0.364230 lr=0.000100 Epoch[026] Batch [0079]/[0149] Speed: 111.103767 samples/sec accuracy=90.566406 loss=0.359991 lr=0.000100 Epoch[026] Batch [0099]/[0149] Speed: 111.428344 samples/sec accuracy=90.750000 loss=0.353162 lr=0.000100 Epoch[026] Batch [0119]/[0149] Speed: 109.321444 samples/sec accuracy=90.794271 loss=0.356022 lr=0.000100 Epoch[026] Batch [0139]/[0149] Speed: 122.987088 samples/sec accuracy=90.904018 loss=0.354871 lr=0.000100 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 026] training: accuracy=90.866191 loss=0.355002 [Epoch 026] speed: 98 samples/sec time cost: 132.306695 [Epoch 026] validation: acc-top1=79.766949 acc-top5=95.153602 loss=0.769064 Epoch[027] Batch [0019]/[0149] Speed: 52.115727 samples/sec accuracy=91.328125 loss=0.362342 lr=0.000100 Epoch[027] Batch [0039]/[0149] Speed: 120.157990 samples/sec accuracy=91.445312 loss=0.344943 lr=0.000100 Epoch[027] Batch [0059]/[0149] Speed: 109.782779 samples/sec accuracy=91.093750 loss=0.350727 lr=0.000100 Epoch[027] Batch [0079]/[0149] Speed: 111.423722 samples/sec accuracy=91.191406 loss=0.347667 lr=0.000100 Epoch[027] Batch [0099]/[0149] Speed: 112.787173 samples/sec accuracy=91.328125 loss=0.343709 lr=0.000100 Epoch[027] Batch [0119]/[0149] Speed: 111.240571 samples/sec accuracy=91.328125 loss=0.342326 lr=0.000100 Epoch[027] Batch [0139]/[0149] Speed: 127.157134 samples/sec accuracy=91.350446 loss=0.341750 lr=0.000100 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 027] training: accuracy=91.296141 loss=0.341281 [Epoch 027] speed: 98 samples/sec time cost: 131.321657 [Epoch 027] validation: acc-top1=79.793432 acc-top5=95.074153 loss=0.762170 Epoch[028] Batch [0019]/[0149] Speed: 50.522042 samples/sec accuracy=90.468750 loss=0.364143 lr=0.000100 Epoch[028] Batch [0039]/[0149] Speed: 117.067648 samples/sec accuracy=91.406250 loss=0.352231 lr=0.000100 Epoch[028] Batch [0059]/[0149] Speed: 105.496774 samples/sec accuracy=91.458333 loss=0.352793 lr=0.000100 Epoch[028] Batch [0079]/[0149] Speed: 112.518770 samples/sec accuracy=91.250000 loss=0.349569 lr=0.000100 Epoch[028] Batch [0099]/[0149] Speed: 106.635747 samples/sec accuracy=91.171875 loss=0.352548 lr=0.000100 Epoch[028] Batch [0119]/[0149] Speed: 111.776425 samples/sec accuracy=90.898438 loss=0.355818 lr=0.000100 Epoch[028] Batch [0139]/[0149] Speed: 127.232248 samples/sec accuracy=90.747768 loss=0.358430 lr=0.000100 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 028] training: accuracy=90.824245 loss=0.359455 [Epoch 028] speed: 96 samples/sec time cost: 134.064872 [Epoch 028] validation: acc-top1=80.190678 acc-top5=95.312500 loss=0.757097 Epoch[029] Batch [0019]/[0149] Speed: 50.979429 samples/sec accuracy=91.328125 loss=0.323999 lr=0.000100 Epoch[029] Batch [0039]/[0149] Speed: 115.523998 samples/sec accuracy=91.406250 loss=0.329580 lr=0.000100 Epoch[029] Batch [0059]/[0149] Speed: 108.562099 samples/sec accuracy=90.937500 loss=0.336404 lr=0.000100 Epoch[029] Batch [0079]/[0149] Speed: 114.004168 samples/sec accuracy=90.566406 loss=0.344459 lr=0.000100 Epoch[029] Batch [0099]/[0149] Speed: 109.105435 samples/sec accuracy=90.500000 loss=0.347243 lr=0.000100 Epoch[029] Batch [0119]/[0149] Speed: 113.843035 samples/sec accuracy=90.598958 loss=0.350518 lr=0.000100 Epoch[029] Batch [0139]/[0149] Speed: 128.490231 samples/sec accuracy=90.591518 loss=0.350366 lr=0.000100 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 029] training: accuracy=90.562081 loss=0.350345 [Epoch 029] speed: 97 samples/sec time cost: 132.383807 [Epoch 029] validation: acc-top1=79.766949 acc-top5=95.127119 loss=0.767295 Epoch[030] Batch [0019]/[0149] Speed: 50.225033 samples/sec accuracy=92.187500 loss=0.343870 lr=0.000100 Epoch[030] Batch [0039]/[0149] Speed: 117.745020 samples/sec accuracy=91.718750 loss=0.345508 lr=0.000100 Epoch[030] Batch [0059]/[0149] Speed: 106.880557 samples/sec accuracy=91.822917 loss=0.342462 lr=0.000100 Epoch[030] Batch [0079]/[0149] Speed: 114.188014 samples/sec accuracy=91.562500 loss=0.341089 lr=0.000100 Epoch[030] Batch [0099]/[0149] Speed: 108.437381 samples/sec accuracy=91.625000 loss=0.341405 lr=0.000100 Epoch[030] Batch [0119]/[0149] Speed: 114.851256 samples/sec accuracy=91.679688 loss=0.339996 lr=0.000100 Epoch[030] Batch [0139]/[0149] Speed: 126.140107 samples/sec accuracy=91.540179 loss=0.340050 lr=0.000100 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 030] training: accuracy=91.715604 loss=0.335790 [Epoch 030] speed: 97 samples/sec time cost: 132.579863 [Epoch 030] validation: acc-top1=80.323093 acc-top5=95.233051 loss=0.752026 Epoch[031] Batch [0019]/[0149] Speed: 50.182249 samples/sec accuracy=91.484375 loss=0.333101 lr=0.000100 Epoch[031] Batch [0039]/[0149] Speed: 116.775856 samples/sec accuracy=90.976562 loss=0.351314 lr=0.000100 Epoch[031] Batch [0059]/[0149] Speed: 107.339960 samples/sec accuracy=91.458333 loss=0.340049 lr=0.000100 Epoch[031] Batch [0079]/[0149] Speed: 113.645317 samples/sec accuracy=91.601562 loss=0.341481 lr=0.000100 Epoch[031] Batch [0099]/[0149] Speed: 107.231042 samples/sec accuracy=91.578125 loss=0.339466 lr=0.000100 Epoch[031] Batch [0119]/[0149] Speed: 112.737126 samples/sec accuracy=91.627604 loss=0.339537 lr=0.000100 Epoch[031] Batch [0139]/[0149] Speed: 126.275284 samples/sec accuracy=91.484375 loss=0.339165 lr=0.000100 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 031] training: accuracy=91.474413 loss=0.340372 [Epoch 031] speed: 97 samples/sec time cost: 133.324835 [Epoch 031] validation: acc-top1=80.084746 acc-top5=95.286017 loss=0.756305 Epoch[032] Batch [0019]/[0149] Speed: 50.557525 samples/sec accuracy=91.796875 loss=0.357746 lr=0.000100 Epoch[032] Batch [0039]/[0149] Speed: 116.413530 samples/sec accuracy=92.695312 loss=0.320582 lr=0.000100 Epoch[032] Batch [0059]/[0149] Speed: 105.472324 samples/sec accuracy=92.630208 loss=0.321878 lr=0.000100 Epoch[032] Batch [0079]/[0149] Speed: 112.529121 samples/sec accuracy=92.558594 loss=0.317675 lr=0.000100 Epoch[032] Batch [0099]/[0149] Speed: 109.225691 samples/sec accuracy=92.312500 loss=0.326611 lr=0.000100 Epoch[032] Batch [0119]/[0149] Speed: 111.785979 samples/sec accuracy=92.161458 loss=0.329898 lr=0.000100 Epoch[032] Batch [0139]/[0149] Speed: 127.253392 samples/sec accuracy=92.087054 loss=0.329544 lr=0.000100 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 032] training: accuracy=92.040688 loss=0.332114 [Epoch 032] speed: 96 samples/sec time cost: 133.608474 [Epoch 032] validation: acc-top1=79.608051 acc-top5=95.180085 loss=0.773102 Epoch[033] Batch [0019]/[0149] Speed: 50.706426 samples/sec accuracy=92.187500 loss=0.317172 lr=0.000100 Epoch[033] Batch [0039]/[0149] Speed: 114.907433 samples/sec accuracy=92.148438 loss=0.321741 lr=0.000100 Epoch[033] Batch [0059]/[0149] Speed: 108.382018 samples/sec accuracy=92.083333 loss=0.328425 lr=0.000100 Epoch[033] Batch [0079]/[0149] Speed: 113.037732 samples/sec accuracy=92.128906 loss=0.321476 lr=0.000100 Epoch[033] Batch [0099]/[0149] Speed: 107.135686 samples/sec accuracy=92.031250 loss=0.324320 lr=0.000100 Epoch[033] Batch [0119]/[0149] Speed: 113.134264 samples/sec accuracy=92.005208 loss=0.322758 lr=0.000100 Epoch[033] Batch [0139]/[0149] Speed: 129.066001 samples/sec accuracy=91.975446 loss=0.322761 lr=0.000100 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 033] training: accuracy=91.956795 loss=0.322445 [Epoch 033] speed: 97 samples/sec time cost: 133.611440 [Epoch 033] validation: acc-top1=79.740466 acc-top5=95.312500 loss=0.766052 Epoch[034] Batch [0019]/[0149] Speed: 50.126714 samples/sec accuracy=92.109375 loss=0.308279 lr=0.000100 Epoch[034] Batch [0039]/[0149] Speed: 114.871764 samples/sec accuracy=91.679688 loss=0.321810 lr=0.000100 Epoch[034] Batch [0059]/[0149] Speed: 107.093680 samples/sec accuracy=91.458333 loss=0.332994 lr=0.000100 Epoch[034] Batch [0079]/[0149] Speed: 114.756530 samples/sec accuracy=91.542969 loss=0.330202 lr=0.000100 Epoch[034] Batch [0099]/[0149] Speed: 104.805545 samples/sec accuracy=91.593750 loss=0.330700 lr=0.000100 Epoch[034] Batch [0119]/[0149] Speed: 113.817826 samples/sec accuracy=91.705729 loss=0.330021 lr=0.000100 Epoch[034] Batch [0139]/[0149] Speed: 124.963759 samples/sec accuracy=91.662946 loss=0.330323 lr=0.000100 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 034] training: accuracy=91.642198 loss=0.330819 [Epoch 034] speed: 96 samples/sec time cost: 133.819625 [Epoch 034] validation: acc-top1=79.581568 acc-top5=95.286017 loss=0.766592 Epoch[035] Batch [0019]/[0149] Speed: 49.883294 samples/sec accuracy=91.406250 loss=0.346315 lr=0.000100 Epoch[035] Batch [0039]/[0149] Speed: 113.752448 samples/sec accuracy=91.718750 loss=0.329638 lr=0.000100 Epoch[035] Batch [0059]/[0149] Speed: 106.610250 samples/sec accuracy=91.901042 loss=0.329031 lr=0.000100 Epoch[035] Batch [0079]/[0149] Speed: 111.742847 samples/sec accuracy=91.914062 loss=0.330365 lr=0.000100 Epoch[035] Batch [0099]/[0149] Speed: 106.608884 samples/sec accuracy=91.781250 loss=0.330788 lr=0.000100 Epoch[035] Batch [0119]/[0149] Speed: 115.090992 samples/sec accuracy=91.809896 loss=0.328709 lr=0.000100 Epoch[035] Batch [0139]/[0149] Speed: 129.029655 samples/sec accuracy=91.662946 loss=0.328545 lr=0.000100 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 035] training: accuracy=91.684144 loss=0.328190 [Epoch 035] speed: 96 samples/sec time cost: 133.582545 [Epoch 035] validation: acc-top1=79.661017 acc-top5=95.233051 loss=0.756305 Epoch[036] Batch [0019]/[0149] Speed: 51.157968 samples/sec accuracy=89.921875 loss=0.351279 lr=0.000100 Epoch[036] Batch [0039]/[0149] Speed: 115.695849 samples/sec accuracy=90.820312 loss=0.339687 lr=0.000100 Epoch[036] Batch [0059]/[0149] Speed: 106.830501 samples/sec accuracy=91.328125 loss=0.328459 lr=0.000100 Epoch[036] Batch [0079]/[0149] Speed: 115.226299 samples/sec accuracy=90.976562 loss=0.335546 lr=0.000100 Epoch[036] Batch [0099]/[0149] Speed: 106.995453 samples/sec accuracy=91.000000 loss=0.334740 lr=0.000100 Epoch[036] Batch [0119]/[0149] Speed: 115.948535 samples/sec accuracy=90.989583 loss=0.336981 lr=0.000100 Epoch[036] Batch [0139]/[0149] Speed: 125.977930 samples/sec accuracy=91.049107 loss=0.336545 lr=0.000100 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 036] training: accuracy=91.013003 loss=0.338181 [Epoch 036] speed: 97 samples/sec time cost: 132.289274 [Epoch 036] validation: acc-top1=80.084746 acc-top5=95.550847 loss=0.749448 Epoch[037] Batch [0019]/[0149] Speed: 49.889188 samples/sec accuracy=91.406250 loss=0.326860 lr=0.000100 Epoch[037] Batch [0039]/[0149] Speed: 115.943324 samples/sec accuracy=91.796875 loss=0.322310 lr=0.000100 Epoch[037] Batch [0059]/[0149] Speed: 105.656266 samples/sec accuracy=91.927083 loss=0.325623 lr=0.000100 Epoch[037] Batch [0079]/[0149] Speed: 115.920350 samples/sec accuracy=91.855469 loss=0.328084 lr=0.000100 Epoch[037] Batch [0099]/[0149] Speed: 106.038886 samples/sec accuracy=91.656250 loss=0.328181 lr=0.000100 Epoch[037] Batch [0119]/[0149] Speed: 113.204402 samples/sec accuracy=91.640625 loss=0.328588 lr=0.000100 Epoch[037] Batch [0139]/[0149] Speed: 127.479275 samples/sec accuracy=91.651786 loss=0.327559 lr=0.000100 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 037] training: accuracy=91.537332 loss=0.330108 [Epoch 037] speed: 96 samples/sec time cost: 134.296110 [Epoch 037] validation: acc-top1=80.190678 acc-top5=95.206568 loss=0.757485 Epoch[038] Batch [0019]/[0149] Speed: 48.881623 samples/sec accuracy=93.125000 loss=0.282256 lr=0.000100 Epoch[038] Batch [0039]/[0149] Speed: 117.274367 samples/sec accuracy=92.929688 loss=0.288081 lr=0.000100 Epoch[038] Batch [0059]/[0149] Speed: 104.722749 samples/sec accuracy=92.994792 loss=0.288994 lr=0.000100 Epoch[038] Batch [0079]/[0149] Speed: 112.830572 samples/sec accuracy=92.324219 loss=0.305082 lr=0.000100 Epoch[038] Batch [0099]/[0149] Speed: 105.492233 samples/sec accuracy=92.171875 loss=0.309233 lr=0.000100 Epoch[038] Batch [0119]/[0149] Speed: 114.893212 samples/sec accuracy=92.135417 loss=0.310742 lr=0.000100 Epoch[038] Batch [0139]/[0149] Speed: 123.603773 samples/sec accuracy=91.997768 loss=0.311553 lr=0.000100 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 038] training: accuracy=92.135067 loss=0.307789 [Epoch 038] speed: 95 samples/sec time cost: 134.227345 [Epoch 038] validation: acc-top1=80.137712 acc-top5=95.391949 loss=0.755862 Epoch[039] Batch [0019]/[0149] Speed: 49.112168 samples/sec accuracy=91.562500 loss=0.318046 lr=0.000100 Epoch[039] Batch [0039]/[0149] Speed: 115.684932 samples/sec accuracy=91.562500 loss=0.317869 lr=0.000100 Epoch[039] Batch [0059]/[0149] Speed: 104.367265 samples/sec accuracy=91.145833 loss=0.328451 lr=0.000100 Epoch[039] Batch [0079]/[0149] Speed: 113.985344 samples/sec accuracy=91.289062 loss=0.323017 lr=0.000100 Epoch[039] Batch [0099]/[0149] Speed: 104.957285 samples/sec accuracy=91.546875 loss=0.320347 lr=0.000100 Epoch[039] Batch [0119]/[0149] Speed: 113.496343 samples/sec accuracy=91.770833 loss=0.319584 lr=0.000100 Epoch[039] Batch [0139]/[0149] Speed: 125.921556 samples/sec accuracy=91.741071 loss=0.318520 lr=0.000100 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 039] training: accuracy=91.883389 loss=0.314846 [Epoch 039] speed: 95 samples/sec time cost: 134.250321 [Epoch 039] validation: acc-top1=80.031780 acc-top5=95.444915 loss=0.758044 Epoch[040] Batch [0019]/[0149] Speed: 48.731852 samples/sec accuracy=91.562500 loss=0.310527 lr=0.000010 Epoch[040] Batch [0039]/[0149] Speed: 114.136964 samples/sec accuracy=92.070312 loss=0.309538 lr=0.000010 Epoch[040] Batch [0059]/[0149] Speed: 105.167468 samples/sec accuracy=92.005208 loss=0.316752 lr=0.000010 Epoch[040] Batch [0079]/[0149] Speed: 113.893417 samples/sec accuracy=91.972656 loss=0.319874 lr=0.000010 Epoch[040] Batch [0099]/[0149] Speed: 106.111064 samples/sec accuracy=91.890625 loss=0.317156 lr=0.000010 Epoch[040] Batch [0119]/[0149] Speed: 113.477991 samples/sec accuracy=91.966146 loss=0.317023 lr=0.000010 Epoch[040] Batch [0139]/[0149] Speed: 128.247766 samples/sec accuracy=91.964286 loss=0.314378 lr=0.000010 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 040] training: accuracy=91.956795 loss=0.314153 [Epoch 040] speed: 95 samples/sec time cost: 134.810678 [Epoch 040] validation: acc-top1=79.978814 acc-top5=95.233051 loss=0.769158 Epoch[041] Batch [0019]/[0149] Speed: 48.639120 samples/sec accuracy=91.328125 loss=0.328120 lr=0.000010 Epoch[041] Batch [0039]/[0149] Speed: 116.116584 samples/sec accuracy=91.562500 loss=0.330849 lr=0.000010 Epoch[041] Batch [0059]/[0149] Speed: 107.502396 samples/sec accuracy=92.161458 loss=0.317121 lr=0.000010 Epoch[041] Batch [0079]/[0149] Speed: 113.447966 samples/sec accuracy=92.246094 loss=0.315144 lr=0.000010 Epoch[041] Batch [0099]/[0149] Speed: 106.424288 samples/sec accuracy=92.296875 loss=0.312557 lr=0.000010 Epoch[041] Batch [0119]/[0149] Speed: 113.402000 samples/sec accuracy=92.213542 loss=0.312791 lr=0.000010 Epoch[041] Batch [0139]/[0149] Speed: 127.239720 samples/sec accuracy=92.232143 loss=0.311306 lr=0.000010 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 041] training: accuracy=92.239933 loss=0.311811 [Epoch 041] speed: 96 samples/sec time cost: 134.351120 [Epoch 041] validation: acc-top1=80.296610 acc-top5=95.206568 loss=0.764379 Epoch[042] Batch [0019]/[0149] Speed: 49.446068 samples/sec accuracy=92.421875 loss=0.294264 lr=0.000010 Epoch[042] Batch [0039]/[0149] Speed: 119.175016 samples/sec accuracy=91.835938 loss=0.306781 lr=0.000010 Epoch[042] Batch [0059]/[0149] Speed: 102.907321 samples/sec accuracy=91.718750 loss=0.308019 lr=0.000010 Epoch[042] Batch [0079]/[0149] Speed: 115.873364 samples/sec accuracy=92.265625 loss=0.297548 lr=0.000010 Epoch[042] Batch [0099]/[0149] Speed: 107.692029 samples/sec accuracy=92.484375 loss=0.290939 lr=0.000010 Epoch[042] Batch [0119]/[0149] Speed: 113.147743 samples/sec accuracy=92.447917 loss=0.292870 lr=0.000010 Epoch[042] Batch [0139]/[0149] Speed: 127.826980 samples/sec accuracy=92.444196 loss=0.295665 lr=0.000010 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 042] training: accuracy=92.512584 loss=0.294124 [Epoch 042] speed: 96 samples/sec time cost: 132.933156 [Epoch 042] validation: acc-top1=80.376059 acc-top5=95.233051 loss=0.755707 Epoch[043] Batch [0019]/[0149] Speed: 48.144010 samples/sec accuracy=91.875000 loss=0.319197 lr=0.000010 Epoch[043] Batch [0039]/[0149] Speed: 117.660199 samples/sec accuracy=92.109375 loss=0.322466 lr=0.000010 Epoch[043] Batch [0059]/[0149] Speed: 106.002380 samples/sec accuracy=92.031250 loss=0.323514 lr=0.000010 Epoch[043] Batch [0079]/[0149] Speed: 113.106956 samples/sec accuracy=92.011719 loss=0.319853 lr=0.000010 Epoch[043] Batch [0099]/[0149] Speed: 102.968757 samples/sec accuracy=92.234375 loss=0.311233 lr=0.000010 Epoch[043] Batch [0119]/[0149] Speed: 114.560278 samples/sec accuracy=92.252604 loss=0.310165 lr=0.000010 Epoch[043] Batch [0139]/[0149] Speed: 128.935833 samples/sec accuracy=92.209821 loss=0.310426 lr=0.000010 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 043] training: accuracy=92.208473 loss=0.311196 [Epoch 043] speed: 95 samples/sec time cost: 134.801189 [Epoch 043] validation: acc-top1=79.899364 acc-top5=95.524364 loss=0.751990 Epoch[044] Batch [0019]/[0149] Speed: 49.992793 samples/sec accuracy=92.812500 loss=0.309715 lr=0.000010 Epoch[044] Batch [0039]/[0149] Speed: 116.241158 samples/sec accuracy=91.992188 loss=0.323524 lr=0.000010 Epoch[044] Batch [0059]/[0149] Speed: 105.054234 samples/sec accuracy=92.317708 loss=0.319164 lr=0.000010 Epoch[044] Batch [0079]/[0149] Speed: 115.186477 samples/sec accuracy=92.285156 loss=0.315500 lr=0.000010 Epoch[044] Batch [0099]/[0149] Speed: 105.004079 samples/sec accuracy=92.390625 loss=0.312043 lr=0.000010 Epoch[044] Batch [0119]/[0149] Speed: 113.266947 samples/sec accuracy=92.382812 loss=0.311430 lr=0.000010 Epoch[044] Batch [0139]/[0149] Speed: 127.895575 samples/sec accuracy=92.332589 loss=0.309495 lr=0.000010 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 044] training: accuracy=92.376258 loss=0.306424 [Epoch 044] speed: 96 samples/sec time cost: 133.879301 [Epoch 044] validation: acc-top1=79.819915 acc-top5=95.206568 loss=0.765172 Epoch[045] Batch [0019]/[0149] Speed: 47.706285 samples/sec accuracy=92.109375 loss=0.316687 lr=0.000010 Epoch[045] Batch [0039]/[0149] Speed: 115.801455 samples/sec accuracy=91.835938 loss=0.310975 lr=0.000010 Epoch[045] Batch [0059]/[0149] Speed: 106.753836 samples/sec accuracy=92.057292 loss=0.308501 lr=0.000010 Epoch[045] Batch [0079]/[0149] Speed: 115.878113 samples/sec accuracy=92.265625 loss=0.310929 lr=0.000010 Epoch[045] Batch [0099]/[0149] Speed: 103.870168 samples/sec accuracy=92.578125 loss=0.303254 lr=0.000010 Epoch[045] Batch [0119]/[0149] Speed: 112.999413 samples/sec accuracy=92.526042 loss=0.304450 lr=0.000010 Epoch[045] Batch [0139]/[0149] Speed: 127.872454 samples/sec accuracy=92.611607 loss=0.305752 lr=0.000010 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 045] training: accuracy=92.512584 loss=0.308055 [Epoch 045] speed: 95 samples/sec time cost: 134.537682 [Epoch 045] validation: acc-top1=80.031780 acc-top5=95.021186 loss=0.769003 Epoch[046] Batch [0019]/[0149] Speed: 48.757753 samples/sec accuracy=92.187500 loss=0.319017 lr=0.000010 Epoch[046] Batch [0039]/[0149] Speed: 116.243770 samples/sec accuracy=91.914062 loss=0.323436 lr=0.000010 Epoch[046] Batch [0059]/[0149] Speed: 108.549546 samples/sec accuracy=92.317708 loss=0.314410 lr=0.000010 Epoch[046] Batch [0079]/[0149] Speed: 115.795778 samples/sec accuracy=92.578125 loss=0.310140 lr=0.000010 Epoch[046] Batch [0099]/[0149] Speed: 105.784206 samples/sec accuracy=92.531250 loss=0.311820 lr=0.000010 Epoch[046] Batch [0119]/[0149] Speed: 115.398885 samples/sec accuracy=92.421875 loss=0.313733 lr=0.000010 Epoch[046] Batch [0139]/[0149] Speed: 127.941326 samples/sec accuracy=92.187500 loss=0.321403 lr=0.000010 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 046] training: accuracy=92.082634 loss=0.322255 [Epoch 046] speed: 96 samples/sec time cost: 132.647184 [Epoch 046] validation: acc-top1=79.978814 acc-top5=95.338983 loss=0.754546 Epoch[047] Batch [0019]/[0149] Speed: 48.916939 samples/sec accuracy=92.265625 loss=0.303852 lr=0.000010 Epoch[047] Batch [0039]/[0149] Speed: 114.245598 samples/sec accuracy=92.187500 loss=0.307580 lr=0.000010 Epoch[047] Batch [0059]/[0149] Speed: 107.999689 samples/sec accuracy=92.109375 loss=0.310275 lr=0.000010 Epoch[047] Batch [0079]/[0149] Speed: 114.577478 samples/sec accuracy=92.167969 loss=0.313670 lr=0.000010 Epoch[047] Batch [0099]/[0149] Speed: 107.071850 samples/sec accuracy=92.515625 loss=0.302665 lr=0.000010 Epoch[047] Batch [0119]/[0149] Speed: 112.648198 samples/sec accuracy=92.239583 loss=0.308590 lr=0.000010 Epoch[047] Batch [0139]/[0149] Speed: 129.034834 samples/sec accuracy=92.276786 loss=0.307389 lr=0.000010 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 047] training: accuracy=92.292366 loss=0.308224 [Epoch 047] speed: 96 samples/sec time cost: 134.185957 [Epoch 047] validation: acc-top1=79.872881 acc-top5=95.312500 loss=0.751525 Epoch[048] Batch [0019]/[0149] Speed: 51.292378 samples/sec accuracy=91.406250 loss=0.325338 lr=0.000010 Epoch[048] Batch [0039]/[0149] Speed: 113.694852 samples/sec accuracy=91.796875 loss=0.326592 lr=0.000010 Epoch[048] Batch [0059]/[0149] Speed: 110.474515 samples/sec accuracy=92.343750 loss=0.315067 lr=0.000010 Epoch[048] Batch [0079]/[0149] Speed: 113.037999 samples/sec accuracy=92.246094 loss=0.314525 lr=0.000010 Epoch[048] Batch [0099]/[0149] Speed: 105.964070 samples/sec accuracy=92.343750 loss=0.312698 lr=0.000010 Epoch[048] Batch [0119]/[0149] Speed: 109.360678 samples/sec accuracy=92.161458 loss=0.316222 lr=0.000010 Epoch[048] Batch [0139]/[0149] Speed: 127.316493 samples/sec accuracy=92.354911 loss=0.311930 lr=0.000010 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 048] training: accuracy=92.439178 loss=0.311466 [Epoch 048] speed: 96 samples/sec time cost: 133.212127 [Epoch 048] validation: acc-top1=80.376059 acc-top5=95.153602 loss=0.741093 Epoch[049] Batch [0019]/[0149] Speed: 50.229421 samples/sec accuracy=91.953125 loss=0.331876 lr=0.000010 Epoch[049] Batch [0039]/[0149] Speed: 116.681177 samples/sec accuracy=92.226562 loss=0.319218 lr=0.000010 Epoch[049] Batch [0059]/[0149] Speed: 106.246610 samples/sec accuracy=92.291667 loss=0.315284 lr=0.000010 Epoch[049] Batch [0079]/[0149] Speed: 111.885493 samples/sec accuracy=92.363281 loss=0.316882 lr=0.000010 Epoch[049] Batch [0099]/[0149] Speed: 107.039767 samples/sec accuracy=92.421875 loss=0.316534 lr=0.000010 Epoch[049] Batch [0119]/[0149] Speed: 114.079914 samples/sec accuracy=92.408854 loss=0.314397 lr=0.000010 Epoch[049] Batch [0139]/[0149] Speed: 128.905621 samples/sec accuracy=92.377232 loss=0.312784 lr=0.000010 Batch [0019]/[0059]: evaluated Batch [0039]/[0059]: evaluated [Epoch 049] training: accuracy=92.407718 loss=0.311077 [Epoch 049] speed: 97 samples/sec time cost: 133.245542 [Epoch 049] validation: acc-top1=80.270127 acc-top5=95.444915 loss=0.744282