Hugging Face's logo

microsoft
/
Phi-4-multimodal-instruct

Automatic Speech Recognition
Transformers
Safetensors
phi4mm
text-generation
nlp
code
audio
speech-summarization
speech-translation
visual-question-answering
phi-4-multimodal
phi
phi-4-mini
custom_code
Eval Results
Model card Files
xet
 Community
94

Decoding strategy of the Phi4 Multimodal

#50
by Zhengyang - opened
Discussion
Zhengyang

Dear authors,
thank you for the great work. What is the decoding strategy of the phi4 multimodal? Is it beam search or topk sampling? I didn't find it in the configuration file.

Best,
Zhengyang

fanruchao

Hi @Zhengyang ,

For speech/audio tasks, we simply used greedy search (top-1) for the benchmark. You can try other options for more diverse output if you like.

Thanks,
Ruchao

Sign up or log in to comment