search

Championchess / A-Tri-Layer-Plugin-to-Improve-Occluded-Detection

Apache License 2.0
97 stars 8 forks source link

The download link is invalid #1

Closed Yuhuoo closed 1 year ago

Yuhuoo commented 2 years ago

Thanks to your nice work, but I cannot download the .pkl using the download link. The download link is invalid. Can you help me?

Championchess commented 2 years ago

Thank you for your interest in our work. You can try to wget it using command line, and it should work.

Yuhuoo commented 2 years ago

Thank you for your interest in our work. You can try to wget it using command line, and it should work.

Ok! Thanks for your reply, and I have another questions. What the LR can I set with 2 GPU and batch_size=4 during training ? How can I calcuate the LR ? Can I use the auto_scale_lr to set LR automatically? The auto_scale_lr is deleted from the default_runtime.py in the project

Championchess commented 1 year ago

As mentioned in the paper, for the fine-tuning process, we set the initial learning rate to be 0.000001 with batch_size=16. I have not tried to do with batch_size=4.