Support for Stable Diffusion image generators

[kreely]

Since these llava files contain.. everything needed for using Wizard Coder, Rocket, etc....

How possible would it be to create a llava file that contained the needed model files and the Python interpreter and libraries to run something like Stable Diffusion.... or image generation packages similar to Stable Diffusion??

This would be pretty amazing, though putting an image generator in the hands of anyone able to download a single cross-platform executable does make me pause a moment.... it could be used quite unethically. There is a similar risk with Open Source tools and these LLM executables anyway..... Thinking out loud.

[jart]

It'd be nice to have an easier way to generate cat photos on the command line. One project we could use is https://github.com/leejet/stable-diffusion.cpp They appear to depend on GGML but they don't have GPU support yet leejet/stable-diffusion.cpp#6 It's only been in recent months that the CLIP code for LLaVA in llama.cpp gained GPU accelleration. Hey @FSSRepo what would you recommend for us, if we want image generation on llama.cpp w/ gpu?

[FSSRepo]

I also intended to use that stable diffusion project with flame.cpp and create something like ChatGPT + DALLE-3, but the truth is that the prompt generation quality of the models I've tried leaves much to be desired. Also, let me clarify that stable diffusion.cpp has GPU support; please read the README

[liar666]

Hi,
Thanks for the work.

I've seen that there are "Stable Diffusion" and "Whisper" subdirs in the code:
https://github.com/Mozilla-Ocho/llamafile/tree/main/stable-diffusion.cpp
https://github.com/Mozilla-Ocho/llamafile/tree/main/whisper.cpp

But there's no documentation (neither in main README nor in subdirs README) on how to make a .llamafile of them.

Is it possible? How?