Modelpedia
Work in progress
About
Findings
Models
Concepts
Methods
Datasets
Sources
Light
Dark
Can We Talk Models Into Seeing the World Differently?
2025-01-22
· ICLR 2025 Poster ·
anchor
Findings
IC-415
VLMs show a default shape bias (47.9-73.8%) that exceeds their vision encoders and vision-only models but falls short of human levels (96%), with the LLM component rather than the encoder responsible for suppressing one visual cue.
IC-416
Natural language prompts can steer the texture/shape bias in VLMs in both directions without significantly affecting accuracy, with texture-biased prompts more effective than shape-biased ones; this steering also generalizes to low/high-frequency bias.