Thanks for sharing the code!
How do you generate image-specific reasons and generic class reasoning using the LMM?
Is get_image_reason_feat.py the correct script for that?
If so, how do you create the .json files (the ones inside ./reason_text) required to run this script?
Thanks for sharing the code!
How do you generate image-specific reasons and generic class reasoning using the LMM?
Is get_image_reason_feat.py the correct script for that?
If so, how do you create the .json files (the ones inside ./reason_text) required to run this script?