Metraipolis › archive › dc_ad3a3ff91fec
TraceRouter++: Causal Semantic Routing Realignment via Optimal Transport — arXiv:2611.04878
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2611.04878
- License (as archived)
- CC0-1.0
- Archived
- 2026-11-04T18:38:44.726Z
- SHA-256 of the archived text
81187cfff3b55dec02d82e01ced36d872cc4e327b86e2a43d440f0c06e7a8ead- Stable URL of this page
https://metraipolis.com/documento/dc_ad3a3ff91fec
How to cite
TraceRouter++: Causal Semantic Routing Realignment via Optimal Transport — arXiv:2611.04878 — arXiv (en), https://arxiv.org/abs/2611.04878 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_ad3a3ff91fec (Metraipolis archive dc_ad3a3ff91fec, sha256 81187cfff3b55dec02d82e01ced36d872cc4e327b86e2a43d440f0c06e7a8ead)
BibTeX
@misc{dcad3a3ff91fec,
title = {TraceRouter++: Causal Semantic Routing Realignment via Optimal Transport — arXiv:2611.04878},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2611.04878},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_ad3a3ff91fec; sha256 81187cfff3b55dec02d82e01ced36d872cc4e327b86e2a43d440f0c06e7a8ead},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.04878. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Despite their remarkable capabilities, large foundation models (LFMs) remain susceptible to adversarial manipulation. Current safety intervention methods primarily rely on the \textit{locality hypothesis}, rigidly suppressing isolated neurons or features. However, harmful semantics manifest as distributed, cross-layer circuits, rendering localized ablations both brittle against complex jailbreaks and detrimental to general utility. To bridge this gap, we propose TraceRouter++, a routing-level safety intervention framework that explicitly traces safety-relevant causal routing pathways activated under harmful inputs and realigns their unsafe activation states. Our framework operates in a \textit{Discover-Trace-Realign} paradigm: it first pinpoints a sensitive onset layer via attention divergence and disentangles dense activations into interpretable routing features using sparse autoencoders (SAEs). Subsequently, it maps these features to downstream causal circuits via feature influence scores (FIS). Crucially, departing from rigid suppression paradigms, TraceRouter++ introduces a constrained optimal transport (OT)-based soft routing realignment. By formulating safety intervention as a tractable coordinate-wise marginal optimal transport approximation, our method redirects unsafe activation marginals toward safe-reference marginals within causally identified routing subspaces while preserving non-selected coordinates at the intervention point. This mechanism safely redirects malicious semantic flows while leaving orthogonal computational routes intact. Extensive experiments across representative large language models (LLMs), multimodal large language models (MLLMs), and diffusion models (DMs) show that TraceRouter++ outperforms evaluated baselines, improving adversarial robustness while maintaining a favorable safety-utility trade-off. WARNING: This paper contains unsafe model responses.
Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.CL, nuovi lavori del giorno).
Categorie arXiv: Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2611.04878 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-11-04T18:32:52.162Z dalla pagina https://arxiv.org/list/cs.CL/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).