Can not reproduce the result about dsbn HOT 1 OPEN

Thank you for your interest in my paper. It's been a while since I worked on it, so I don't remember much, but it seems like the commands were correctly written.
Firstly, since the PyTorch version is old, there might be slightly different results if not reverted to the previous version. Even the CUDA version can affect the training. Additionally, it's important to check whether the parameters of the ResNet50 ImageNet pre-trained model provided by PyTorch match those of the previous version. Also, the values reported in the paper are averages from multiple random seeds, so I recommend running them on different random seeds for validation.

from dsbn.