Llama-3 Instruct tokenizer_config.json changes in relation to the currently fetched llama-bpe configs.

[Spacellary]

Prerequisites

Please answer the following questions for yourself before submitting an issue.

• I am running the latest code. Development is very rapid so there are no tagged versions as of now.
• I carefully followed the README.md.
• I searched using keywords relevant to my issue to make sure that I am creating a new issue that is not already open (or closed).
• I reviewed the Discussions, and have a new bug or useful enhancement to share.

---

Question/Conjecture:

I am performing model conversions as per the guidelines in this PR and using the llama-bpe configs fetched:

#6920 (comment)

...

The recent convert-hf-to-gguf-update.py script fetches the llama-bpe configs, but these reflect the ones from the Base model.

Recently, within the last week, there was a change to these settings in the meta-llama/Meta-Llama-3-8B-Instruct repo.

Is this change in the Instruct EOS pertinent to the current conversion process?

To add:
I haven't noticed any issues so far using either the Base model configs or the Instruct model configs.

[github-actions]

This issue was closed because it has been inactive for 14 days since being marked as stale.

[Spacellary]

🤗