fastllm

fastllm is a high-performance large model reasoning library implemented in pure C++ without third-party dependencies

6~7B level models can also run smoothly on the Android terminal

Visit Official Website

https://github.com/ztxz16/fastllm