Metraipolis › archive › dc_97ba3f21140d
CaLR: Causal Latent Revision for Robust Diffusion Reasoning — arXiv:2609.20981
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.20981
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-10T01:01:34.835Z
- SHA-256 of the archived text
9b2d4dbdc7c83e33ea3a5f2c323e87c34b7e01a4482275ba6c206b1f654537a7- Stable URL of this page
https://metraipolis.com/documento/dc_97ba3f21140d
How to cite
CaLR: Causal Latent Revision for Robust Diffusion Reasoning — arXiv:2609.20981 — arXiv (en), https://arxiv.org/abs/2609.20981 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_97ba3f21140d (Metraipolis archive dc_97ba3f21140d, sha256 9b2d4dbdc7c83e33ea3a5f2c323e87c34b7e01a4482275ba6c206b1f654537a7)
BibTeX
@misc{dc97ba3f21140d,
title = {CaLR: Causal Latent Revision for Robust Diffusion Reasoning — arXiv:2609.20981},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.20981},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_97ba3f21140d; sha256 9b2d4dbdc7c83e33ea3a5f2c323e87c34b7e01a4482275ba6c206b1f654537a7},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.20981. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Autoregressive (AR) models suffer from local greediness, while diffusion language models (DLMs) often lack the strict causal structure required for reasoning. To combine the advantages and overcome the drawbacks of the dual, we propose Causal Latent Revision (CaLR), a framework that reformulates reasoning as constrained latent optimization. By adopting a causal topology matrix (CTM) from an expert model and implicit differentiation, CaLR performs gradient-guided ``thought revision" to enforce logical consistency, enabling dynamic self-correction of intermediate steps during parallel generation. Empirically, CaLR achieves SOTA DLM performance on complex benchmarks, surpassing strong AR baselines and demonstrating superior robustness in constrained tasks like Sudoku. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-21 in cs.AI). Categorie arXiv: Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2609.20981 — 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 2026-10-10T00:23:42.425Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-21?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).