Is there a official script for inference or finetuning with audio modality?
#50
by
JasonLee996
- opened
As title. Meanwhile, I find it said in README at github repo that the model could be tuned with Align-Anything, but in the funetune.py in offical repo, MiniCPM-o/finetune/finetune.py set "init_audio=False", so will it be OK if I use this script to tuning the audio pathway with this parameter "True"?
We have supported audio modality fine-tuning for MiniCPM-o on LLaMA-Factory. You can refer to the LLaMA-Factory documentation for processing audio datasets.😊