ARES is a novel framework that systematically discovers and mitigates dual vulnerabilities in RLHF systems by simultaneously testing the core LLM and its Reward Model (RM) using structured adversarial…

ConMoE proposes a train-free method for compressing Mixture-of-Experts (MoE) models by consolidating the large expert pool into a smaller set of reusable prototypes and deterministically remapping all…

The paper proposes a local perturbation theory showing that cross-domain interference in multi-domain RL occurs via a low-dimensional shared conflict subspace, which can be selectively mitigated by sh…

RogueMerge introduces a unified framework to robustly attack LLM model merging by addressing the challenges of autoregressive decoding, unknown merging configurations, and prompt generalization, signi…

dMoE proposes a block-level Mixture-of-Experts (MoE) framework for Diffusion Large Language Models (dLLMs) that aggregates token-level expert distributions into a unified block-level distribution, sig…

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…

The paper introduces TrojanMerge, a framework demonstrating that model merging can be exploited to systematically compromise the safety alignment of multiple individually safe LLMs.

MASCing is a novel framework that enables flexible, non-retraining reconfiguration of Mixture-of-Experts (MoE) models for specific safety objectives by applying activation steering masks to control ex…

SkillRevise is an execution-grounded framework that iteratively refines initial, imperfect LLM agent skills by diagnosing defects from execution evidence and applying empirically validated edits, sign…

The paper introduces RA-MoE, a novel fine-tuning framework that leverages the internal routing structure of Mixture-of-Experts (MoE) models to improve performance on multilingual downstream tasks by a…

The paper proposes ReuseRL, a method that improves agent generalization in Reinforcement Learning by enforcing structural compressibility of successful agent trajectories into reusable skills.

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…

CARE-RL introduces a framework combining protocol-aware reward generation and capability-aware optimization to effectively mitigate cross-domain conflicts in multi-domain reinforcement learning for LL…

The paper demonstrates that jointly training a single lightweight neural reranker on multiple diverse environments significantly improves action selection performance and achieves positive cross-domai…

The paper proposes DAG-MoE, a novel sparse Mixture-of-Experts framework that replaces standard weighted-sum aggregation with structural aggregation to enhance model performance and enable multi-step r…

ReSkill is an RL-in-the-loop framework that reconciles skill creation and policy optimization by automatically creating, testing, and refining modular skills alongside the agent's policy learning, lea…

The paper proposes Resonant Context Anchoring (RCA), a lightweight, training-free method that enhances factual faithfulness in LLMs by dynamically amplifying the signal of external context evidence du…

The paper introduces an AI red teaming agent that drastically reduces the time and effort required for security testing by allowing operators to define complex attack goals using natural language, com…

This paper introduces Anchored Weight Decay (AWD), a regularization technique that effectively prevents prior-task forgetting during LLM fine-tuning with Evolution Strategies (ES), positioning ES as a…

The paper introduces Residualized Sparse Autoencoders (ReSAEs) to improve multi-layer interventions in transformers by training each layer on the residual activation, which better preserves cross-laye…