Ask HN: HotPin – lossless 120B MoE inference on 24GB RAM (CPU, 50 loc)

The HotPin project demonstrates lossless inference of a 120B MoE model using only 24GB RAM and CPU with just 50 lines of code, offering a new approach for edge deployment.

• 开源
• 推理优化
• MoE
• HotPin
• CPU