Repository navigation

[TRTLLM-13117][feat] Implement Uneven TP Linear for VisualGen models - #14875

Merged

belgarten-nv merged 6 commits into

NVIDIA:mainfrom

belgarten-nv:user/belgarten/uneven-tp

Jul 21, 2026

Merged

[TRTLLM-13117][feat] Implement Uneven TP Linear for VisualGen models#14875
belgarten-nv merged 6 commits into
NVIDIA:mainfrom
belgarten-nv:user/belgarten/uneven-tp

Commits on Jul 10, 2026
- uneven tp initial impl
  belgarten-nv
  committed
  fe9fce1
- fix coderabbit comments
  belgarten-nv
  committed
  4ffb804
- address Shreyas' comments, add lpips test
  belgarten-nv
  committed
  33a4d7b
- fix ci: fused scale loading, auto sharding with skip_create_weights
  belgarten-nv
  committed
  e6af38c
- rebase + address Yiyun's comments
  belgarten-nv
  committed
  9187418
Commits on Jul 20, 2026
- Merge branch 'main' into user/belgarten/uneven-tp
  zhenhuaw-me
  authored
  cc5526b