ConceptAI & ML
Minimal change benchmark
A dataset where images and captions differ only in one targeted aspect such as object, attribute, counting, or spatial relation to test fine-grained understanding.
Know first
Named as a prerequisite by the lectures that define this. The number is how many say so. A grey one has no page yet.
Image Captioning1Vision-language models1
Lecturers
1Connections
2The lectures that define it or assume it, and the concepts next to it.