Commit Graph

3 Commits

Author SHA1 Message Date
qhy
aba2a90045 算子融合 2026-02-07 16:40:33 +08:00
cb334f308b DDIM loop 内小张量分配优化,attention mask 缓存到 GPU,加速30s左右 2026-01-18 22:37:55 +08:00
25c6fc04db 第一次完整测例跑完 2026-01-18 00:30:10 +08:00