[Chore]: TTS/omni: use `vllm.vllm_flash_attn` instead of `flash-attn` package for 5 models

April 25, 2026 · #3131
View on GitHub
Python Difficulty: Medium

Parent Repository

vllm-project/vllm-omni

Python repository

4,517 844
All Issues Back to vllm-omni

Sign in required

Authenticate to use favourites & bookmarks

5