Project-MONAI
Perceptual Loss errors out after hitting the maximum number of downloads
Describe the bug Currently, models are hosted in Google Drive and once the number of downloads hits the limit, the networks cannot be pulled.
To Reproduce Steps to reproduce the behavior: Depending on number of downloads, errors show up when instancing perceptual loss.
Expected behavior No error. We have discussed porting these networks (python files defining them and artifacts) to Huggingface or similar.
Additional context This has already been discussed internally.
This is specifically related to https://github.com/warvito/MedicalNet-models which is a dependency of the loss function. This repository uses gdrive to store weights, we need to move this to huggingface or other storage.
Sounds good to me! Please, feel free to host a copy of the code or/and model parameters in other locations (like Huggingface).
Great thanks @Warvito! Our objective now is to move the repository to one within Project MONAI and host weights on Huggingface. We also want to combine models from https://github.com/Warvito/radimagenet-models into the repository as well. These are both your public repos, I hope combining them isn't a problem (they're both under Apache 2 anyway).