Repo access request rejected?

#65
by deleted - opened
deleted

I requested access to the Llama 3.2 repo and it was rejected. Access was granted through the Meta website... but I'm finding that the hugging face script to convert Meta's model checkpoint files to hf format using convert_llama_weights_to_hf.py, because "11B" is not a supported model_size, as in, it's not in the dict called NUM_SHARDS.
(1) Why would I have been immediately rejected from this repo? I literally just created a hf account to access these models in hf format, or so I thought, and if it's open access on the meta site, why not here?
(2) Is there another way to convert Meta's version of llama 3.2 11B vision to hf format?

Sign up or log in to comment