search

DavidMChan / caption-by-committee

Using LLMs and pre-trained caption models for super-human performance on image captioning.
Other
40 stars 4 forks source link

update object extraction prompts for lower variation #7

Closed anishk23733 closed 1 year ago

anishk23733 commented 1 year ago

Added stricter rules to prevent object extractor from extracting objects that might not be in the image/relevant for the task.

DavidMChan commented 1 year ago

Hi @anishk23733 - can you resolve the merge conflicts? Otherwise LGTM.