Tinted Frames: Question Framing Blinds Vision-Language Models (opens in new tab)

(davidhalladay.github.io)

4 pointsritwikgupta6d ago1 comments

1 comments

VLMs are selectively blind — they decide how much to look at an image based on question framing (open-ended vs Yes/No vs MCQ), even when the same visual reasoning is required.

j / k navigate · click thread line to collapse

Tinted Frames: Question Framing Blinds Vision-Language Models (opens in new tab)

(davidhalladay.github.io)

4 pointsritwikgupta6d ago1 comments

1 comments

ritwikguptaOP6d ago

VLMs are selectively blind — they decide how much to look at an image based on question framing (open-ended vs Yes/No vs MCQ), even when the same visual reasoning is required.

j / k navigate · click thread line to collapse