Pre-compiled llama.cpp

[CMorrison82z]

I'm using an AMD graphics card and had to compile llama.cpp with additional flags and parameters.

From what I can see, llm-chain-llama attempts to compile the llama.cpp module itself, which I imagine is unlikely to produce a suitable result for me.

Is there any way I can use this library with a pre-compiled llama.cpp binary ?

[williamhogman]

If I recall correctly llama.cpp don't provide ready made dynamic libraries to link with so in this case it is hard for us to offer anything along those lines

Any ideas?