Vibe-Eval

Hard visual prompts judged by humans and a model, aimed at day-to-day multimodal usefulness.

Read the original source — github.com

benchmark · Shared by tscosj

0 comments

No comments yet.