The models performed reasonably well on straightforward queries about visual content, like identifying debris on a reef, but struggled significantly w/queries requiring expert knowledge, like IDing specific biological conditions or behaviors.

→ View original post on X — @mit_csail