How to Convert/Quantize Hugging Face Models to GGUF Format | Step-by-Step Guide

Поделиться
HTML-код
  • Опубликовано: 4 янв 2025

Комментарии • 7

  • @am2dan
    @am2dan 3 месяца назад +6

    A couple of differences with the version of llama.cpp as of Sept. 2024:
    - convert.py is replace by convert_hf_to_gguf.py which does not take the --vocab-type argument.
    - quantize is now llama-quantize.
    Otherwise, I think everything else is the same. Thanks for this video,. It clearly presents exactly the info I was looking for.

  • @LBPFrost
    @LBPFrost 6 месяцев назад +2

    actually glad I watched this video.
    simple enough I got it set up twice.
    Cheers mate.
    I had to just run make though instead of curl (cd'd into repo)
    Ubuntu WSL2 windows

    • @DigiDecode
      @DigiDecode  6 месяцев назад

      I am glad you found the video useful, thank you & cheers 👍🏻

  • @aaronchantrill7338
    @aaronchantrill7338 24 дня назад

    I usually just use git with git-lfs to download the whole folder

  • @karavanamektuplar197
    @karavanamektuplar197 3 месяца назад

    Hi. How can I convert a language model from pytorch to ggml on the Huggingface site? There is a tutorial on Ggerganov's github page under the title “convert-pt-to-ggml.py” but I don't fully understand it. I want to convert a Chinese language model to ggml and use it in Subtitle Edit. Can you make a video of this tutorial?