Based on?

#1
by DefamationStation - opened

Hello

Whats this 13b model based on? Llama 3 didn't come with a 13b model is this a merge of some sort? or is it just Llama 2 and it's a typo?

Pruna AI org

Hello, it is based on the following model: MaziyarPanahi/Llama-3-13B-Instruct-v0.1

johnrachwanpruna changed discussion status to closed

cc @MaziyarPanahi the author of hf.co/MaziyarPanahi/Llama-3-13B-Instruct-v0.1

This model is a self-merge of meta-llama/Meta-Llama-3-8B-Instruct model.

Thanks @julien-c , appreciate it.

Also, for those interested, based on this discussion I had made 2 other merges but this time on base model for further fine-tuning: https://huggingface.co/MaziyarPanahi/Llama-3-16B-Instruct-v0.1/discussions/8

Pruna AI org

Thanks @julien-c , appreciate it.

Also, for those interested, based on this discussion I had made 2 other merges but this time on base model for further fine-tuning: https://huggingface.co/MaziyarPanahi/Llama-3-16B-Instruct-v0.1/discussions/8

It looks like the LLama-3-13B link is broken ?

Sorry I fixed it. It's actually 11B, which is what the 16B is coming from https://huggingface.co/MaziyarPanahi/Llama-3-11B

Sign up or log in to comment