Why Limit the Residual Stream to Layers and Not Tokens? Persistent Memory for Continuous Latent Reasoning

• URL: http://arxiv.org/abs/2606.07720v1
• Date: Fri, 05 Jun 2026 15:45:44 GMT
• Title: Why Limit the Residual Stream to Layers and Not Tokens? Persistent Memory for Continuous Latent Reasoning
• Abstract summary: We propose textbfAGCLR (Adaptive Gated Continuous Latent Reasoning), which augments CoCoNuT with a textitGated Concept Stream.<n>AGCLR achieves consistent improvements across all types of datasets.
• Score: 0.10742675209112622
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Large language models (LLMs) have demonstrated remarkable reasoning abilities on mathematical and multi-hop planning tasks. The CoCoNuT (Chain of Continuous Thought) paradigm~\cite{hao2024coconut} extends this by enabling models to reason in latent space, exploring multiple reasoning paths simultaneously rather than committing to a single chain early on. However, we identify a limitation we term the \textbf{concept bottleneck}. At each reasoning pass, intermediate hidden states are overwritten, causing the model to lose critical facts computed in earlier steps as reasoning depth increases. We observe this empirically. On HotpotQA, vanilla CoCoNuT (10.4\% EM) fails to improve over the CoT baseline (11.0\% EM), and performance degrades with curriculum depth on GSM8K. To address this, we propose \textbf{AGCLR} (Adaptive Gated Continuous Latent Reasoning), which augments CoCoNuT with a \textit{Gated Concept Stream}. A persistent residual memory maintained across all reasoning passes, controlled by three learned gates: a \textit{write} gate that commits intermediate facts to memory, a \textit{read} gate that retrieves relevant prior states, and a \textit{forget} gate that prunes irrelevant context. Evaluated on GSM8K, HotpotQA, and ProsQA using GPT-2 as our base model, AGCLR achieves consistent improvements across all types of datasets. With the performance gap compounding as curriculum depth increases, directly resolving the concept bottleneck. Code available at https://anonymous.4open.science/r/JJJJ/README.md

Related papers

• Latent Recurrent Thoughts: Recurrent Refinement of Proposed Latents for Reasoning with Frozen LLMs [16.72007494160683]
  Chain-of-thought reasoning unfolds in discrete token space.<n> Reasoning instead in a model's continuous representation space.<n>We instantiate this as Latent Recurrent Thoughts.
  arXiv  Detail & Related papers  (2026-09-01T11:56:17Z)
• Enhancing Software Engineering Through Closed-Loop Memory Optimization [50.418699819003486]
  We introduce ours, a closed-loop framework for memory augmentation in software engineering (SE) agents.<n>ours grounds memory utility in textitvalidated downstream impact, establishing utility as a task-agnostic textbf benchmark and an annotation-free textbfoptimization signal.<n>Results demonstrate that ours consistently improves SE agents across settings, achieving absolute gains of up to $uparrow5.25%$ in success rate and $uparrow4.63%$ in resolve efficiency.
  arXiv  Detail & Related papers  (2026-06-04T03:17:21Z)
• Stateful Reasoning via Insight Replay [51.85629502016196]
  Chain-of-Thought (CoT) reasoning has become a foundation for eliciting multi-step reasoning in large language models.<n>We identify a major cause of this phenomenon: as the CoT grows, the model's attention to critical insights produced earlier in the trace gradually weakens.<n>We propose textbfInsightReplay, a stateful reasoning approach in which the model periodically extracts critical insights from its reasoning trace and replays them near the active generation frontier.
  arXiv  Detail & Related papers  (2026-05-14T06:52:59Z)
• STOP: Structured On-Policy Pruning of Long-Form Reasoning in Low-Data Regimes [13.293115227628775]
  Long chain-of-thought (Long CoT) reasoning improves performance on multi-step problems, but it also induces overthinking.<n>We propose STOP (Structured On-policy Pruning), an on-policy algorithm for analyzing and pruning long-form reasoning traces.
  arXiv  Detail & Related papers  (2026-05-13T08:28:05Z)
• When to Think, When to Speak: Learning Disclosure Policies for LLM Reasoning [60.714416943748866]
  In single-stream autoregressive interfaces, the same tokens both update the model state and constitute an irreversible public commitment.<n>We introduce Side-by-Side (SxS) Interleaved Reasoning, which makes disclosure timing a controllable decision within standard autoregressive generation.
  arXiv  Detail & Related papers  (2026-05-05T02:59:58Z)
• Rethinking Token Reduction for Large Vision-Language Models [95.48478689025696]
  Large-Language Models (LVLMs) excel in visual understanding and reasoning, but the excessive visual tokens lead to high inference costs.<n>We propose a learning-based prompt-agnostic, termed MetaCompress, overcoming the limitations of generalization designs.
  arXiv  Detail & Related papers  (2026-03-23T08:40:08Z)
• Probing to Refine: Reinforcement Distillation of LLMs via Explanatory Inversion [84.20493238687187]
  We introduce a novel framework that moves beyond simple mimicry to instill a deeper conceptual understanding.<n>underlinetextitFirst, to address pattern memorization, Explanatory Inversion (EI) generates targeted explanatory probes''<n>underlinetextitSecond, to improve generalization, Explanatory GRPO (texttEXGRPO) uses a reinforcement learning algorithm with a novel Dialogue Structure Utility Bonus.
  arXiv  Detail & Related papers  (2026-02-26T23:01:46Z)
• G-MemLLM: Gated Latent Memory Augmentation for Long-Context Reasoning in Large Language Models [6.5798851769146784]
  We propose textbfG-MemLLM, a memory-augmented architecture that integrates a frozen backbone with a trainable textbfLatent Memory Bank.<n>G-MemLLM significantly enhances multi-hop reasoning and relational precision, achieving a 13.3% accuracy boost on ZsRE for Llama 3.1-8B.
  arXiv  Detail & Related papers  (2026-01-12T03:30:29Z)
• SIM-CoT: Supervised Implicit Chain-of-Thought [108.30049193668083]
  Implicit Chain-of-Thought (CoT) methods offer a token-efficient alternative to explicit CoT reasoning in Large Language Models.<n>We identify a core latent instability issue when scaling the computational budget of implicit CoT.<n>We propose SIM-CoT, a plug-and-play training module that introduces step-level supervision to stabilize and enrich the latent reasoning space.
  arXiv  Detail & Related papers  (2025-09-24T17:01:32Z)
• Latent Chain-of-Thought? Decoding the Depth-Recurrent Transformer [0.8738725605667471]
  Chain-of-thought (CoT) reasoning has enabled transformer-based language models to excel at complex mathematics and multi-step planning.<n>In standard decoder-only architectures, these reasoning steps are externalized in natural language, improving interpretability at the cost of efficiency.<n>We investigate whether such reasoning structures emerge in Huginn-3.5B, a depth-recurrent Transformer that reuses layers at inference time without increasing parameter count.
  arXiv  Detail & Related papers  (2025-07-02T23:35:21Z)
• ThinkLess: A Training-Free Inference-Efficient Method for Reducing Reasoning Redundancy [8.962703809086628]
  ThinkLess is an inference-efficient framework that terminates reasoning generation early and maintains output quality without modifying the model.<n>We show that ThinkLess achieves comparable accuracy to full-length Chain-of-Thought (CoT) decoding while greatly reducing decoding time and memory consumption.
  arXiv  Detail & Related papers  (2025-05-21T15:58:16Z)
• Beyond 'Aha!': Toward Systematic Meta-Abilities Alignment in Large Reasoning Models [86.88657425848547]
  Large reasoning models (LRMs) already possess a latent capacity for long chain-of-thought reasoning.<n>We explicitly align models with three meta-abilities: deduction, induction, and abduction, using automatically generated, self-verifiable tasks.<n>Our three stage-pipeline individual alignment, parameter-space merging, and domain-specific reinforcement learning, boosts performance by over 10% relative to instruction-tuned baselines.
  arXiv  Detail & Related papers  (2025-05-15T17:58:33Z)
• Program of Thoughts Prompting: Disentangling Computation from Reasoning for Numerical Reasoning Tasks [108.4568236569645]
  Chain-of-thoughts prompting (CoT) is by far the state-of-art method for these tasks. We propose Program of Thoughts' (PoT), which uses language models to express the reasoning process as a program. PoT can show an average performance gain over CoT by around 12% across all the evaluated datasets.
  arXiv  Detail & Related papers  (2022-11-22T21:06:00Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.