> I would care more about LLaMA architecture when I can get hands on The architecture is available just fine, both in the paper[1] and in code: https://github.

> I would care more about LLaMA architecture when I can get hands on The architecture is available just fine, both in the paper[1] and in code: https://github.com/facebookresearch/llama/blob/main/llama/model.py https://github.com/facebookresearch/llama/blob/main/llama/mo... It's the model weights that you can't get as easily. [1] https://research.facebook.com/publications/llama-open-and-efficient-foundation-language-models/ https://research.facebook.com/publications/llama-open-and-ef...