The paper introduces Metacognitive Memory Policy Optimization (MMPO), a novel memory training approach that optimizes LLM memory not based on final task success, but on minimizing epistemic uncertaint…

The paper introduces a novel framework to quantify faithful confidence expression (FC) in Large Reasoning Models (LRMs), finding that FC remains a significant and challenging reliability target for th…

The paper analyzes the entropy dynamics of Chain-of-Thought (CoT) reasoning, identifying a transition from an exploratory Uncertainty Region to a stable Confidence Region, which enables superior early…

The paper introduces Canonical-Context On-Policy Distillation (CCOPD) to improve multi-turn language model performance by mitigating 'self-anchored drift,' ensuring consistent answers regardless of wh…

The paper demonstrates that extended pure neural reasoning fails on complex, deterministic state-tracking tasks beyond a certain 'Deterministic Horizon,' necessitating the integration of external tool…

The paper introduces a diagnostic testbed, PAVE, to evaluate how LLMs arbitrate between their internal knowledge and retrieved evidence during fact-checking, revealing that this arbitration is unrelia…

The paper proposes COSE, a method that uses an LLM's intrinsic confidence as an uncertainty signal to improve self-evolutionary training, achieving state-of-the-art performance on general reasoning an…

The paper introduces LinTree, a method that explicitly structures the search history of LLM reasoning traces using parent pointers, significantly improving task performance and search efficiency compa…

This paper analyzes failure modes in collaborative visual reasoning systems, demonstrating that naive shared workspaces can amplify hallucinations and proposing diagnostics for improving communication…

This paper synthesizes over 150 scattered studies and reports to provide the first comprehensive primer on post-training reasoning data, organizing the field around data objects, utility, construction…

DREAM-R is a novel framework that significantly enhances speculative reasoning in large multimodal models by optimizing draft generation alignment, introducing a robust verification mechanism, and ena…

MindZero introduces a self-supervised reinforcement learning framework that trains multimodal large language models (MLLMs) for efficient and robust online mental reasoning without requiring explicit…

The paper investigates how language models perform entity tracking across state changes and finds that LMs use a non-incremental, parallel aggregation strategy rather than maintaining a true internal…

DenoiseRL is a novel reinforcement learning framework that improves reasoning in large language models by optimizing directly from the failures and incorrect reasoning traces of weak models, eliminati…

The paper proposes PaW, a co-training framework that uses standard RL rollouts to provide auxiliary world model supervision directly during policy training, significantly improving language agent perf…

The paper identifies and measures a critical failure mode where LLM agents violate policies by losing or corrupting directive-bearing state during the process of assembling the decision context, and p…

LongTraceRL addresses long-context reasoning challenges by generating highly challenging training data and introducing a fine-grained rubric reward, significantly improving evidence-grounded reasoning…

The paper analyzes the failure modes of aggressive 2-bit quantization in large reasoning models, proposing lightweight controls like FP16 planning and loop rescue to restore accuracy and achieve pract…

The paper introduces Abstract Worlds Semantics (AWS), a set-theoretic framework that treats worlds as primitive elements to provide a unified and generalized analysis of various belief change models.

The paper investigates the limits of self-evolution in LLM reasoning under closed-loop settings, finding that while self-improvement is significant, it consistently falls short of perfect oracle super…