Updated Llama 2 to 3 paths

[d-kleine]

• Currently, when storing the pretrained model files for Llama 2 and 3 (3, 3.1, 3.2) for example in llama3-files folder, the text generation model (Meta-Llama-3-8B) will be overwritten by the instruction-finetuned model (Meta-Llama-3-8B-Instruct).
• Therefore, I think it would be better to store the pretrained model files in individual folders, split up by their name on HF.
• I have decided to go with the original naming Meta used on HF, where the naming (e.g., -chat vs. -Instruction) and the capitalization of the billions param letter is inconsistent (e.g., 7b vs. 8B).

[review-notebook-app[bot]]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[d-kleine]

Do you think is is fine or do you want to make any adjustments (e.g. moving the models in subfolders)?

[rasbt]

Thanks for the update!

Do you think is is fine or do you want to make any adjustments (e.g. moving the models in subfolders)?

I'd say simple (no subfolders) is better