- Rating
- 2
- Confidence
- 3
- Informativeness
- 3
- Presentation
- 3
- Soundness
- 2
Read review
Paper summary
This paper studies where perceptual knowledge is transferred in VLMs using perception-side model merging. It finds that primitive cues emerge early, while relation-sensitive information and the strongest functional effects concentrate in middle visual layers.
Critical strengths
- Clear problem formulation. The paper isolates perception-side transfer by keeping the LLM frozen and intervening only on the visual modules.
- Complementary analyses. Linear probing and bidirectional layer replacement provide both representational and functional evidence.
- Consistent layer-wise finding. The results suggest that primitive cues emerge early, while relation-sensitive transfer is concentrated in middle layers. 28294_Tracing_Perceptual_Trans
Critical weaknesses
- Unclear necessity of model merging. Similar layer-wise analyses could potentially be performed by directly comparing the base and perception-tuned models, so the specific role of merging is not fully justified.
- Mean pooling may obscure spatial relations. Averaging all visual tokens can discard localization and pairwise structure, which is especially relevant for relation-sensitive geometric attributes.
- Insufficient justification for the probing design. The paper relies on mean-pooled layer representations and linear classifiers, but does not sufficiently justify why this probing setup is appropriate for capturing complex geometric relations.
- New models and datasets are suggested.
Questions for authors
see above