| 文件 | 最后提交记录 | 最后更新时间 |
|---|---|---|
feat(offload): add host/NUMA shared CPU block offload across model families (#1530) Integrates shared CPU block offload for Hunyuan Image 3.0, MiniMax-H3, Qwen-Image-2512, and Wan 2.1, with host/NUMA weight sharing, simplified launchers, CLI scope selection, and bilingual usage guides. --------- Co-authored-by: liuhongda <liuhongda@sensetime.com> Co-authored-by: helloyongyang <yongyang1030@163.com> | 10 天前 | |
refactor(api)!: unify inference parameters and remove RIFE (#1511) | 18 天前 | |
refactor(api)!: unify inference parameters and remove RIFE (#1511) | 18 天前 |