杨雨豪
Employers on the program
- 2026SGLang1 talk
Events1
Conferences1
Talks1
Years active2026
The year around now
Talks in the past 12 months and announced talks ahead, one row per conference, January to December. Hover a mark for the talk.
Appearances by year
- AICon 全球人工智能开发与应用大会 (AICon Global AI Development and Application Conference)
Talks
| Date | Event | Conference | Role | Title | Links |
|---|---|---|---|---|---|
| Aug 21, 2026 | AICon 全球人工智能开发与应用大会 | 深圳 | AICon 全球人工智能开发与应用大会 (AICon Global AI Development and Application Conference) | speaker | 百万上下文下的 DeepSeek V4:SGLang 推理优化实战 as SGLang, RadixArk Engineer DeepSeek V4 模型采用了由 SWA(滑窗注意力)、CSA(压缩稀疏注意力,4:1 压缩+TopK 筛选)和 HCA(高压缩注意力,128:1 压缩)组成的混合注意力架构,相较于 V3 单一的 MLA 结构更为复杂,同时模型支持百万级上下文长度,对推理框架的缓存管理、算子效率和并行策略提出了严苛挑战。…Show lessDeepSeek V4 模型采用了由 SWA(滑窗注意力)、CSA(压缩稀疏注意力,4:1 压缩+TopK 筛选)和 HCA(高压缩注意力,128:1 压缩)组成的混合注意力架构,相较于 V3 单一的 MLA 结构更为复杂,同时模型支持百万级上下文长度,对推理框架的缓存管理、算子效率和并行策略提出了严苛挑战。 SGLang 通过全栈优化实现了对 DeepSeek V4 的 Day-0 支持。缓存层面,提出 ShadowRadix 机制,为三种注意力分别维护独立的 KV Cache 映射与页表,并通过 Tombstone 自动释放 SWA 过期缓存,配合 … |
Frequent co-speakers
No one else spoke at two or more of the same events.
Speakers with the same name are merged into one entry, so this page may combine more than one person. Affiliations are shown as each conference listed them.