Qwen2.5-Omni-7B-Demo / requirements.txt
aimeri's picture
Update process_input function in app.py to handle audio generation output more robustly, introducing a fallback mechanism for text generation in case of unexpected output formats. Improve error handling during audio and text generation processes. Additionally, update requirements.txt to include flash-attn for enhanced performance.
c98fc82
raw
history blame
196 Bytes
transformers @ git+https://github.com/huggingface/transformers@3a1ead0aabed473eafe527915eea8c197d424356
qwen-omni-utils[decord]
soundfile
torch
gradio
torchvision
torchaudio
accelerate
flash-attn