- Experimental codes are stored under the
script
directory - more details will be listed after the psivt 2023 proceedings
- we can use bash script to train the model, for example of bash script
num_of_TTA=2
js_diverfence=True
python main.py --model_name resnet18 \
--dataset cifar10 \
--save_directory_name resnet18_cifar10_"$num_of_TTA"_"$js_diverfence" \
--milestones '100,150' \
--weight_decay 5e-4 \
--momentum 0.9 \
--e 200\
--num_of_TTA "$num_of_TTA" \
--js_divergence_regularization "$js_diverfence" \
--training_model_name random_pe_resnet18_cifar10_"$num_of_TTA"_"$js_diverfence".t7 \
> main_resnet_cifar10_"$num_of_TTA"_"$js_diverfence".txt