FastLLM

Pure C++ LLM inference library — high performance CPU/GPU, no dependencies.