Optimal placement of perception and cognition layers

Determine whether the perception anchor layer 16 and cognition injection layers 24 and 28 used by Dual-Stream Cross-Anchor Correction are optimal through systematic layer-wise sweeps and sensitivity analyses.

Background

The architecture places the perception anchors at layer 16 and inserts cognition-stream cross-attention at layers 24 and 28. These choices are justified by design considerations about the division between visual perception and language-level reasoning, but the experiments ablate the presence of the streams rather than the specific layer indices.

The paper therefore leaves unresolved whether another selection of perception or cognition layers would produce better grounding or hallucination-mitigation performance, and proposes layer-wise probes and sensitivity analysis as future work.

References

The mechanistic reading that the cognition stream structurally severs the path along which deep reasoning drifts from the visual evidence is confirmed by no rigorous causal analysis and should be read as a structural design intention only; the current choices of perception anchor layer and cognition injection layers follow design considerations, and no layer-by-layer sweep or sensitivity analysis was performed, so this combination cannot be asserted to be optimal; every difference reported here is a point estimate from a single evaluation run without significance testing; and the out-of-domain evaluations use a scoring protocol different from the official leaderboards, string matching on some benchmarks and large-model scoring on others, so their absolute scores are not directly comparable with any leaderboard and the associated conclusions are restricted to the relative ranking of the four configurations under one scorer, with MMHal-Bench returning a null result on a small sample of non-COCO images that serves only as boundary evidence for domain-conditionality.

— Dual-Stream Cross-Anchor Correction Grounding Long-Form Captions and the Domain Limits of Object-Level Anchors  (2608.12746 - Bu, 13 Aug 2026) in Section 6, Conclusion and Future Work; Appendix A.1