Autotrain on Nvidia DGX Cloud not working


After the most recent update, AutoTrain on Nvidia DGX Cloud on HuggingFace no longer functions correctly. When you attempt to start the space and click on the OAuth option for HuggingFace, an Internal Server Error is returned. From my observations, it appears that AutoTrain is reloading, causing the CSRF token to mismatch. Is this a known issue?

Same problem

@Adilmar The issue is that the number of workers on the Dockerfile was changed from 1 to 2. I solved the issue cloning the space locally, changing the number of workers back to 1 and then restarting the space.

1 Like

it should work with 2 workers too! we have now made improvements to oauth just yesterday. lemme know if it doesnt work

1 Like

@vladiacobhackajob @abhishek Got it, thank you very much I restarted my space so far it’s going well.

Kind Regards