Skip to content

Unavailable pre-trained model configs #737

Description

@easygold

Describe the bug
Hi!
I am using finetune library with GPT2 small config and it seems I am not able to download the pre-trained weights anymore.
It seems the required files ['encoder.json', 'vocab.bpe', 'model-sm.jl'] are not accessible.

Minimum Reproducible Example
A short code snippet which reproduces the exception:

from finetune.base_models import GPT2Small
from finetune import Classifier
def main():
    print("Downloading GPT2Small...")
    Classifier(base_model=GPT2Small)

if __name__ == "__main__":
    main()

The web service the library is trying to use to download the files required:

Expected behavior
When I init my GPT2 classifier model I expect the required files ['encoder.json', 'vocab.bpe', 'model-sm.jl'] are downloaded automatically without errors, instead of getting "HTTP Error 403".

Additional context
@benleetownsend it looks like the library is linked to your s3 private bucket "bendropbox"?

Thanks in advance

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions