The paper introduces Weak-Critic Strong Oversight, a method where a weak model guides a strong model's self-improvement by providing non-misleading revision directions, leading to scalable oversight.

The paper introduces a logic-driven framework using a neural certificate function to rigorously evaluate and benchmark the generalization capabilities of reinforcement learning algorithms on unseen ta…

The paper introduces Lookahead Group Reward (&) to combat Supervision Fidelity Decay (SFD) in on-policy distillation, significantly improving student model performance on long reasoning tasks.

The paper introduces Trust-Region behavior Blending (TRB), a warmup method that improves on-policy distillation by replacing poor early student rollouts with teacher-aligned behavior policies, leading…

The paper introduces Trust Region On-Policy Distillation (TrOPD), a robust method that stabilizes the on-policy distillation of large language models by restricting training to regions where teacher s…

The paper proposes Preference Delta Aggregation (PDA), a framework that aggregates multiple weak preference signals derived from smaller model pairs using LoRA merging to significantly boost the perfo…

CRITIC-R1 introduces a structured critic framework that treats RAG critique as an explicit error diagnosis problem using reinforcement learning, significantly improving answer quality over strong RAG…

This paper develops a unified spectral analysis framework to explain how knowledge transfer (KT) works across different machine learning regimes, such as Knowledge Distillation and Weak-to-Strong gene…

The paper proposes Skill-Conditioned Gated Self-Distillation (SGSD), a novel framework that uses retrieved, potentially noisy skills to guide LLM reasoning, achieving state-of-the-art performance on m…

The paper introduces Responsible Contrastive Soft Prompting (RCSP), a parameter-efficient method using soft prompts to improve LLM reliability by simultaneously suppressing hallucinations, encouraging…

The paper proposes a novel structural invariant approach, derived from the economic constraints of fraud, that amplifies weak, low-precision signals into highly accurate fraud detections without requi…

The paper identifies a new class of difficult-to-detect trustworthiness failures, termed 'Silent Failures,' that arise when personalizing foundation models using federated learning, arguing that curre…

The paper introduces 'reward bias substitution,' demonstrating that single-axis mitigations of reward model biases merely shift optimization pressure to correlated proxies, and proposes augmenting eva…

The paper demonstrates that current defenses against malicious fine-tuning of foundation models are insufficient because they only address fixed attacks, and introduces a unified adaptive attack that…

The paper proposes DySCo, a dynamic trust-aware sparse consensus mechanism, to efficiently manage communication in multi-agent LLM systems by selectively connecting agents based on real-time value, th…

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…

This paper provides the first non-vacuous generalization analysis for the Stochastic Variance Reduced Gradient (SVRG) method by establishing sharp, data-dependent algorithmic stability bounds, thereby…

DASH introduces a dual-branch distillation framework to effectively compress class-conditional diffusion models by independently supervising both score branches, significantly preserving guidance fide…

The paper introduces the quotient semivalue mechanism to provide fair data attribution that is resistant to contributors manipulating their reported identities by splitting or duplicating data.

This paper investigates the phenomenon of 'copying' in Distribution Matching Distillation (DMD), finding that high-dimensional distillation causes student models to spontaneously reproduce the teacher…