Metraipolis › archive › dc_bda2bdbf2c53

Latent Recurrent Thoughts: Recurrent Refinement of Proposed Latents for Reasoning with Frozen LLMs — arXiv:2609.01117

Source: arXiv (en)
Original: https://arxiv.org/abs/2609.01117
License (as archived): CC0-1.0
Archived: 2026-10-17T06:32:22.112Z
SHA-256 of the archived text: aad49a7e5cd6b64b992e090fb7dd89d6fbd6d1d06bdffd06ccccad7fa8b16f8e
Stable URL of this page: https://metraipolis.com/documento/dc_bda2bdbf2c53

How to cite

Latent Recurrent Thoughts: Recurrent Refinement of Proposed Latents for Reasoning with Frozen LLMs — arXiv:2609.01117 — arXiv (en), https://arxiv.org/abs/2609.01117 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_bda2bdbf2c53 (Metraipolis archive dc_bda2bdbf2c53, sha256 aad49a7e5cd6b64b992e090fb7dd89d6fbd6d1d06bdffd06ccccad7fa8b16f8e)

BibTeX

@misc{dcbda2bdbf2c53,
  title = {Latent Recurrent Thoughts: Recurrent Refinement of Proposed Latents for Reasoning with Frozen LLMs — arXiv:2609.01117},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.01117},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_bda2bdbf2c53; sha256 aad49a7e5cd6b64b992e090fb7dd89d6fbd6d1d06bdffd06ccccad7fa8b16f8e},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.01117. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Chain-of-thought reasoning unfolds in discrete token space: each step is committed as text, errors propagate, and eliciting good traces presupposes traces to imitate. Reasoning instead in a model's continuous representation space - where intermediate states are vectors rather than words - sidesteps these constraints, but leaves open how those latent states should be computed. We approach this along two axes. First, we keep a large language model (LLM) frozen and use it for what it is already good at - modeling and decoding sequences - while a small auxiliary network supplies continuous latent thoughts as input. Second, we produce those latents by recurrence: a tiny recurrent reasoner refines them over many steps, decoupling the depth of computation from the size of the model, so that the latents are a product of iterative processing rather than a single forward pass. We instantiate this as Latent Recurrent Thoughts (LRT): a task-dedicated proposer supplies base latents, a recurrent reasoner refines them through bounded residual corrections, and the frozen LLM decodes the answer. On symbolic reasoning with answer supervision but no reasoning traces (Countdown-4, Sudoku) and on natural-language reasoning (HumanEval, MBPP, StrategyQA), LRT substantially outperforms prior frozen-decoder continuous-space reasoning methods under an identical decoder, prompt, data, and training budget, and outperforms non-thinking-mode chain-of-thought prompting on the same backbone at a small fraction of its inference compute.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-02 in cs.AI).
Categorie arXiv: Artificial Intelligence (cs.AI) ; Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2609.01117 — 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 2026-10-17T06:15:12.677Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-02?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).