Fastllm: A LLM inference library that runs DeepSeek-V4 with 10GB VRAM

(github.com)

3 points | by nogajun 9 hours ago ago