Hacker News new | ask | show | jobs
by homarp 147 days ago
it'a a new architecture. Not yet implemented in llama.cpp

issue to follow: https://github.com/ggml-org/llama.cpp/issues/18931