Open fabianocastello opened 1 year ago
I am getting the same exact error.
Attempted solution but did not work: -Created write access token on HuggingFace website -Login successfully both on Jupyter Notebook and in terminal with huggingface-cli login command
Have anyone resolved this issue?
I finally found a fix: push_to_hub=False instead of True and it seems to train and save the model locally without the git error
training_args = TrainingArguments(output_dir=model_name, num_train_epochs=2, learning_rate=2e-5, per_device_train_batch_size=batch_size, per_device_eval_batch_size=batch_size, weight_decay=0.01, evaluation_strategy="epoch", disable_tqdm=False, logging_steps=logging_steps, push_to_hub=False, log_level="error")
Another better fix, this solution will allow you to upload to the HuggingFace ML
-Make sure your parent directory is NOT a Git, if you clone the book repo, it is an initiated Git repo. Therefore, you have to delete .git folder manually in the transformer book repo -Ensure that the dir at output_dir inside TrainingArguments is an empty directory -Install the latest transformers pip install transformers==4.31.0
This error simply means that the repository that should be cloned onto your account was not found. To resolve this, log in to HuggingFace, create a new repository by clicking on "+ New Model", and the issue should be resolved.
Information
The question or comment is about chapter:
Question or comment
I am trying to run the notebook in Colab. In cell:
I got an error stating the 'https://huggingface.co/fcastell/distilbert-base-uncased-finetuned-emotion/' was not found. Could be something relate to git lfs but I try to "git clone" manually and the root cause seems do be that the repository http is not valid.
What am I missing? Thanks for help.
This is the entire error log: