reimplement classic CNN model and test on CIFAR-10 dataset
- python 3.5+
- pytorch 1.0+
This project is released under the MIT license
Please Choose the model you would like to use at train.py
, and using the model
defined in the <classic-CNN-model dir>/models
replace the variable model
in the train.py
after you choose the model, just use
python train.py --lr <learning rate> --resume <resume_path> --gpu <whether use gpu> --work_dir <model save directory> --max_epoch<max_epoch>
can start the training process