llamafile icon indicating copy to clipboard operation
llamafile copied to clipboard

GUI for creating llamafiles

Open faceradix opened this issue 1 year ago • 5 comments

I read all detail. But I do not understand how I make from .gguf to .llamafile Please let me know. Because I am not developer. Thank you

faceradix avatar Dec 09 '23 17:12 faceradix

Thanks for reaching out. I'm going to file this issue as a feature request for either a web GUI or a desktop GUI that makes it easier to create llamafiles. While we're happy to consider your request, I'd be surprised if there's many non-technical people out there who are interested in publishing models. We'll see what @stlhood thinks.

jart avatar Dec 09 '23 19:12 jart

문의해 주셔서 감사합니다. 이 문제는 라마 파일을 더 쉽게 생성할 수 있는 웹 GUI 또는 데스크톱 GUI에 대한 기능 요청으로 제출하겠습니다. 우리는 귀하의 요청을 기꺼이 고려하지만, 모델 게시에 관심이 있는 비기술적인 사람들이 많다면 놀랄 것입니다. 우리는 무엇을 볼 것이다@stlhood생각한다. Other models on my old MacBook are so slow, and I'm so happy to run into llamafile. I'm an undeveloped person, but I want to create an automation program with language models. Thank you so much.

faceradix avatar Dec 09 '23 19:12 faceradix

@faceradix Are you wanting to run a GGUF file with llamafile?

If so then you don't need to create a llamafile for that GGUF. Rather you can download the llamafile-server file from the releases page

And then call it from your command line, passing in the GGUF model as an argument:

./llamafile-server-0.2.1 -m my-gguf-file.gguf

mneedham avatar Dec 10 '23 20:12 mneedham

@faceradix Are you wanting to run a GGUF file with llamafile?

If so then you don't need to create a llamafile for that GGUF. Rather you can download the llamafile-server file from the releases page

And then call it from your command line, passing in the GGUF model as an argument:

./llamafile-server-0.2.1 -m my-gguf-file.gguf

oh! it's working!! really really thank you so much :)

faceradix avatar Dec 12 '23 07:12 faceradix

It's working too! Thanks. But now I wonder if I can compile it and make it a single executable file without always calling it from my cmd.

DjagbleyEmmanuel avatar Dec 15 '23 13:12 DjagbleyEmmanuel

I've just merged #112 which lets you create llamafiles with a single simple command. Assuming you make && sudo make install the llamafile repo beforehand, then all you have to do is say llamafile-convert URL where you give it some GGUF weights URL off Hugging Face, and then a llamafile will pop out. I think that's honestly better than a GUI and you can thank @chand1012 for contributing it.

jart avatar Jan 02 '24 02:01 jart