Commit Graph

5 Commits (184a65370451452bae87d0058bba06563028c4a8)

Author SHA1 Message Date
Steve Luo 7806842f2d
add paged-attetionv2: support seq length split across thread block (#5707)
7 months ago
傅剑寒 50104ab340
[Inference/Feat] Add convert_fp8 op for fp8 test in the future (#5706)
7 months ago
Steve Luo 5cd75ce4c7
[Inference/Kernel] refactor kvcache manager and rotary_embedding and kvcache_memcpy oper… (#5663)
7 months ago
Steve Luo a8fd3b0342
[Inference/Kernel] Optimize paged attention: Refactor key cache layout (#5643)
7 months ago
傅剑寒 279300dc5f
[Inference/Refactor] Refactor compilation mechanism and unified multi hw (#5613)
7 months ago