Metraipolis › archive › dc_ad292dde2586
Steering Recurrent Reasoners at Inference Time with Readout Feedback — arXiv:2608.24136
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2608.24136
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-18T06:36:05.482Z
- SHA-256 of the archived text
9c0b03a88f5d607f40004755ec37aa498c8d67792596f2623ea21427cc53c910- Stable URL of this page
https://metraipolis.com/documento/dc_ad292dde2586
How to cite
Steering Recurrent Reasoners at Inference Time with Readout Feedback — arXiv:2608.24136 — arXiv (en), https://arxiv.org/abs/2608.24136 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_ad292dde2586 (Metraipolis archive dc_ad292dde2586, sha256 9c0b03a88f5d607f40004755ec37aa498c8d67792596f2623ea21427cc53c910)
BibTeX
@misc{dcad292dde2586,
title = {Steering Recurrent Reasoners at Inference Time with Readout Feedback — arXiv:2608.24136},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2608.24136},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_ad292dde2586; sha256 9c0b03a88f5d607f40004755ec37aa498c8d67792596f2623ea21427cc53c910},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.24136. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Recurrent models, which repeatedly update latent states with shared computation blocks, have emerged as powerful architectures for solving complex reasoning tasks. Existing inference-time methods scale computation by running more steps or sampling more trajectories, but ignore information revealed within each trajectory. Here we show that recurrent models can be improved at inference time by using their own readout probabilities to steer latent dynamics without retraining. We introduce Readout Feedback (RoFB), a test-time intervention that converts intermediate predictions into token-wise pairwise coupling forces injected into the latent dynamics. Across three recurrent models (AKOrN, ItrSA++, TRM) on Sudoku and Maze, RoFB yields clear gains in four of six model-task pairs and a small gain in one. The four positive pairs achieve performance unattainable by merely running more steps or selecting from multiple trajectories, at comparable or lower computational cost. These results suggest that closed-loop steering of latent dynamics can serve as a complementary inference-time control mechanism for recurrent reasoning models. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-26 in cs.LG). Categorie arXiv: Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2608.24136 — 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 2026-10-18T06:28:52.230Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-08-26?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).