arXiv:2606. 02871v1 Announce Type: cross Abstract: Large reasoning models improve performance by generating extended chain-of-thought (CoT) reasoning, but this behavior becomes inefficient when applied to LLM agents.
arXiv:2505.16782v3 Announce Type: replace Abstract: Large Language Models (LLMs) have shown impressive performance on complex tasks through Chain-of-Thought (CoT) reasoning. However, conventional CoT...
arXiv:2511. 20639v3 Announce Type: replace-cross Abstract: Multi-agent systems (MAS) extend large language models (LLMs) from independent single-model reasoning to coordinative system-level intelligence.
arXiv:2511. 16886v5 Announce Type: replace-cross Abstract: Recently, small models with latent recursion have obtained promising results on complex reasoning tasks.
arXiv:2510. 15416v2 Announce Type: replace Abstract: We investigate a framework in which LoRA adapters are treated as callable tools that a base language model can dynamically select and invoke.
arXiv:2606. 16360v1 Announce Type: cross Abstract: Chain-of-thought (CoT) prompting improves reasoning in large language models (LLMs) by externalizing intermediate computation as discrete text tokens, but this textual interface also introduces redundancy and inference overhead.
arXiv:2601. 03093v2 Announce Type: replace Abstract: Recent work on activation and latent steering has demonstrated that modifying internal representations can effectively guide large language models (LLMs) toward improved reasoning and efficiency without updating model parameters.
arXiv:2606. 16222v1 Announce Type: new Abstract: Large Language Models (LLMs) increasingly rely on intermediate reasoning, yet explicit Chain-of-Thought (CoT) suffers from a linguistic space bottleneck: each thought must be decoded into tokens, causing high inference overhead.
The paper argues that large language models need adaptive reasoning rather than fixed reasoning budgets. It shows that over‑reasoning leads to high computational cost without accuracy gains, while under‑reasoning results in incorrect or incomplete solutions. The authors evaluate these failure modes on MATH‑500 and the GAIA benchmark, highlighting the need for dynamic reasoning allocation in agentic AI systems.
The paper introduces a method to improve test-time scaling (TTS) for large language models by using multi-agent systems (MAS) to split long reasoning chains into manageable contexts. A new dataset, M500, containing 500 multi-agent collaborative reasoning traces, is used to fine‑tune open‑source models, enabling them to learn collaborative patterns and outperform their base versions. An adaptive scaling strategy with a "CEO" agent is proposed to dynamically guide reasoning depth, and experiments in the AgentVerse framework confirm the effectiveness of the approach.
The paper introduces a text-based, multi-agent version of the board game Clue to test multi-step deductive reasoning in large language models (LLMs). Six LLM-based agents (GPT‑4o‑mini and Gemini‑2.5‑Flash) play turn‑based games, and a tool‑augmented approach uses a structured possibility matrix to convert implicit game state into explicit remaining possibilities, thereby offloading memory and deductive constraints from the agents. The study compares this tool‑augmented method against a baseline to assess its impact on reasoning quality and task success in a strategic reasoning environment.
arXiv:2606. 03965v1 Announce Type: cross Abstract: Large language models improve final-answer accuracy through extended chain-of-thought reasoning, but often spend tokens inefficiently and offer little inference-time control.
LatentHarness unifies memory access and latent reasoning by treating them as sequential latent actions—THINK, RECALL, and EXIT—within a language model. It is trained via counterfactual policy distillation, which evaluates the impact of each action on the emitted token and learns when to recall evidence versus continue reasoning. On six long‑context reasoning benchmarks, a 1.4B‑parameter LatentHarness model outperforms the strongest baselines by 2.8% and 10.0% relative, while running 5.9× faster than the leading long‑context baseline.