I followed course on Question answering - Hugging Face Course and even though it creates a folder in huggingface but the upload is not happening. Please not i logged in using “write API token”
AFAIK this should’ve been fixed. Will try to reproduce your error 
its getting stored locally… under the name of the folder supplied in currect directory
It’s normal that it’s stored locally, what callback does is that after every step/epoch it saves and pushes your model asynchronously, the pushing part here is apparently not working. Your internet might be the cause, the model size might be another reason, or something might be up (which I’m currently checking 
)
A temporary fix is that after your model is trained you can just push the model itself with git-lfs.
pls share a link for this… i have tried a few and they dont work on colab
Even local storage seems to have stopped working here… there is no update after epoch 1 even though model is running 4th epoch
Doing model.push_to_hub() should work for any model In the link you will find the documentation to this.
I think we’ve seen few times issues with colab and mounted drive causing extremely slow uploads Trainer.push_to_hub is taking lot of time, is this expected behaviour?
2 Likes
@vanichandna care to share the colab notebook you’re working out of if you’re still having problems? That way we may be able to assist a bit better.
here are all the issues i m facing
- no auto uploading of models on huggingface
- training never stops… even after all the epochs complete, training cell doesnt stop running.
- no intermediate updation of epochs locally