Machine Learning Reproducibility Challenge (MLRC)

Prepare Dataset (CIFAR-10)

cd data
wget https://www.cs.toronto.edu/~kriz/cifar-10-python.tar.gz

Run Test : valid.sh

model=[model name] (e.g., resnet20, resnet32, resnet44, resnet56)
pretrained_model=[weight url] (e.g., ./save_lr_001_resnet56/checkpoint.th)
echo "python -u validation.py  --arch=$model  --pretrained=$pretrained_model" 
python -u validation.py  --arch=$model  --pretrained=$pretrained_model

weight configuration detail

save_cosine_[model name] : Using Cosine Scheduler
save_lr_001_[model name] : Learning Rate Tuning
save_[model_name]base : Base Line Model
save[model_name]_nores : Residual Abalation Study

Train Detail (LOG File)

log_cosine_[model name] : Using Cosine Scheduler
log_lr_001_[model name] : Learning Rate Tuning
log_[model_name]base : Base Line Model
log[model_name]_nores : Residual Abalation Study