color_syncnet_train.py training not working

sahilg06 / EmoGen

PyTorch Implementation for Paper "Emotionally Enhanced Talking Face Generation" (ICCVW'23 and ACM-MMW'23)

Creative Commons Zero v1.0 Universal

343 stars 27 forks source link

color_syncnet_train.py training not working #13

Closed charlie89517 closed 1 year ago

charlie89517 commented 1 year ago

After executing the command "python color_syncnet_train.py --data_root preprocessed_dataset/ --checkpoint_dir C:\EmoGen\checkpoints\syncnet," my CPU usage stays at 100% for several hours, but it seems that the training has not started yet. I'm not sure at which stage I'm stuck.

charlie89517 commented 1 year ago

To let others know, I found the issue, and it was related to the version of librosa. I updated it to version 0.9.0, and it started working correctly.

sarthak268 commented 1 year ago

Thank you @charlie89517 for the clarification.

charlie89517 commented 1 year ago

I also noticed that when the librosa version reaches 0.10.1, there may be issues. I recommend keeping the librosa version at 0.9.1.

sarthak268 commented 1 year ago

Done. Thank you!