ggerganov/llama.cpp Released Version b11765
llama.cpp release b11765 includes minor model router optimizations, specifically dropping redundant precision overrides in the Kolibri-1 router. This routine point release helps streamline model execution paths and code maintenance, contributing to the overall stability of local LLM inference tools. The update removes two ggml_prec_set_* calls since the router matrix multiplication (matmul) functions correctly using the default precision.
## BACKGROUND
llama.cpp is an open-source software library developed in C/C++ that enables efficient local inference on various large language models using the GGML tensor library. It serves as a foundational core for numerous local AI applications and inference tools across multiple operating systems and hardware backends.