Metraipolis › archive › dc_908bcaa43bcd
Residual Streams Read, Recurrent States Remember: The Global Workspace in Mamba Models — arXiv:2609.32102
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.32102
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-07T20:55:07.863Z
- SHA-256 of the archived text
716f006adb95085094f255fb5d965083791d0970c70d6a59a908ef06eed59051- Stable URL of this page
https://metraipolis.com/documento/dc_908bcaa43bcd
How to cite
Residual Streams Read, Recurrent States Remember: The Global Workspace in Mamba Models — arXiv:2609.32102 — arXiv (en), https://arxiv.org/abs/2609.32102 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_908bcaa43bcd (Metraipolis archive dc_908bcaa43bcd, sha256 716f006adb95085094f255fb5d965083791d0970c70d6a59a908ef06eed59051)
BibTeX
@misc{dc908bcaa43bcd,
title = {Residual Streams Read, Recurrent States Remember: The Global Workspace in Mamba Models — arXiv:2609.32102},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.32102},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_908bcaa43bcd; sha256 716f006adb95085094f255fb5d965083791d0970c70d6a59a908ef06eed59051},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.32102. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Can the global-workspace account of transformer representations extend to state-space language models? We fit Jacobian lenses to the residual streams and recurrent states of Mamba-1, Mamba-2 and Mamba-3, using the original 1000-prompt recipe. Joint residual--state readouts improve recovery of known intermediate concepts over the residual lens on at least five of six task families in every tested Mamba checkpoint. On Mamba-2, state alone exceeds residual and logit lenses on all six families; a normalised joint readout improves on both components on five. Temporal maps and word-list experiments show earlier content remaining state-readable as residual visibility changes. We also propose sign-guarded steering, which improves target top-five success over coordinate exchange on matched verbal-report trials in five models. Recurrent state alone supports this verbal access. These gains do not extend consistently to relational answers: guarded edits often output the edited concept itself, and Mamba-3's joint edits can disrupt successful state-only redirection. Recurrent state thus provides a complementary carrier of workspace content, whose recovery, persistence and causal uses require separate measurements. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-29 in cs.AI). Categorie arXiv: Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2609.32102 — 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 2026-10-07T20:29:00.709Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).