Repository navigation

[Feature][Operator] Add DeepSeek V4.1 sparse attention operators - #16422

Merged

weijinqian0 merged 7 commits into

vllm-project:mainfrom

GDzhu01:codex/deepseek-v4.1-operators

Sep 14, 2026

Commits on Sep 12, 2026
- [Feature][Operator] Add DeepSeek V4.1 sparse attention operators
  GDzhu01
  committed
  caa11e6
- [Docs] Fix Aurora compilation matrix link
  GDzhu01
  committed
  fd87bf4
- [Docs] Avoid unresolved operator source link
  GDzhu01
  committed
  e53d0b0
- [Fix] Preserve hc pre v2 operator compatibility
  GDzhu01
  committed
  e791a85
- [Fix] Preserve QLI metadata device argument
  GDzhu01
  committed
  db04d32
- Preserve quant lightning indexer v2 compatibility
  GDzhu01
  committed
  99aa88c
- Route legacy QLI calls through extended backend
  GDzhu01
  committed
  001bb52