tau

joined 2 years ago
[โ€“] [email protected] 2 points 9 months ago (1 children)

Do you have any tips (or examples) using quadlets? I tried using them but I couldn't wrap my head around them.

[โ€“] [email protected] 4 points 9 months ago

Just beware that like AMD, Intel GPUs suffer a performance hit when using LLMs because of the CUDA specific optimizations in frameworks like llama.cpp