Better emdings support

[hrstoyanov]

Not fan of ONNX .. is there a way to use llama.cpp and and have something built-in? See https://github.com/tobi/qmd

[Goldziher]

Interesting.

I'll take a look.

Can you elaborate? Or others as well, on the embedding needs and what's currently not good?

[hrstoyanov]

I noticed that QMD does not require ONNX runtime or anything extra, just works out-of-the-box, something I liked.

This is just my preference, maybe not important for others

[hrstoyanov]

Nevermind what I asked, I am closing this