Transformers
GGUF
English
mistral
Generated from Trainer
text-generation-inference

Will there be a re-upload of this model?

#3
by SolidSnacke - opened

I would just like to know about this. I saw that there was an update of the files for this model, although I’ll say right away that I don’t understand anything about this and perhaps it was stupid on my part to ask this question.

No, I don't see any reason to re-upload.

I checked the source repo and you're right, there was an update. But it didn't change the model. Rather, they additionally uploaded the base model in a different format - pytorch_model.bin. It was previously uploaded only in a new format, called model.safetensors. The config.json also changed, but that was only because they had used a more recent version of Transformers to create it, so the transformers version number changed. That's purely informational, and doesn't change anything about how the model operates.

I don't know why they did the new upload in the pytorch_model.bin format, as that format is becoming deprecated. But I can see that the original model.safetensors files are unchanged. So the version I quantised is still up-to-date.

I also don't think they'd do a model update without making a new repo, and giving it a new name and/or version number.

You can carry on using these GGUF, they're up-to-date with the source Zephyr repo.

Thank you for your reply.

Sign up or log in to comment