vllm
[Bugfix] Keep legacy WNA16 MoE dispatch on Triton under batch invariance
#61038
Merged

Commits
  • [Bugfix] Keep legacy WNA16 MoE dispatch on Triton under batch invariance
    yifanFengg committed 40 days ago
  • Update tests/kernels/moe/test_moe.py
    yifanFengg committed 35 days ago
  • Drop the comment in should_moe_wna16_use_cuda
    yifanFengg committed 35 days ago
  • Merge branch 'main' into bugfix/wna16-moe-batch-invariant
    vllm-agent committed 25 days ago
Loading