run_se_inception_resnet_v2.sh

#!/usr/bin/env sh
export MXNET_CUDNN_AUTOTUNE_DEFAULT=1

# you'd better change setting with your own --data-dir, --depth, --batch-size, --gpus.

## tgrain se-inception-v4
# LD_PRELOAD="/usr/lib/libtcmalloc.so" python -u train_se_inception_resnet_v2.py --data-dir data/imagenet --data-type imagenet --batch-size 384 --gpus=6,7,8,9
python -u train_se_inception_resnet_v2.py --data-dir data/imagenet --data-type imagenet --batch-size 384 --gpus=6,7,8,9