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

• Posted 3 hours ago by nogajun
• 3 points