AyUsir / SENet-pytorch

Implementing SENet using the pytorch framework,You can use tensorboard to view the log and resume training at checkpoints

SENet-pytorch

Implementing SENet using the pytorch framework

requirement

• python3.8
• pytorch1.6
• torchvision
• tensorboard
• tqdm

dataset

The data set is automatically downloaded to the data folder, if the speed is too slow,You can try the following link https://share.weiyun.com/56FKfYz password：nwdmtc

train

python train.py

If you want to train a model without SE block

python train.py --baseline

Trainers can change parameters in train.py according to their needs.

log

The log file will be saved in the logs folder

 tensorboard --logdir=logs

You can view the training process and results

some result

model	ResNet20	SE-ResNet20
max val accuracy	92.2%	92.5%