939 papers
AlphaApollo is an agentic reasoning system that enhances foundation models' performance on complex, long-horizon tasks by orchestrating multi-turn agentic reasoning, turn-level reinforcement learning for tool-use optimization, and a propose-judge-update evolution loop with verification.
The paper proposes MSSR, a memory-aware adaptive replay framework that estimates sample-level memory strength to dynamically schedule rehearsal intervals, effectively mitigating catastrophic forgetting while maintaining fast adaptation in continual LLM fine-tuning.
The paper introduces MITRA, an on-premise Retrieval-Augmented Generation (RAG) system designed to enhance knowledge retrieval in large-scale physics collaborations like CMS by employing a novel automated pipeline for document processing and a two-tiered vector database architecture to accurately answer context-aware questions while ensuring data privacy.
This paper introduces EXPLORE-Bench, a benchmark derived from real first-person videos to evaluate the ability of multimodal large language models to perform long-horizon egocentric scene prediction, revealing significant performance gaps compared to humans and demonstrating that stepwise reasoning offers partial improvements at a computational cost.
This paper introduces RbtAct, a framework that leverages peer review rebuttals as implicit supervision to train large language models to generate more actionable and specific review feedback through a novel perspective-conditioned task and a new dataset called RMR-75K.
This paper introduces MUGEN, a comprehensive benchmark revealing that Large Audio-Language Models struggle with multi-audio understanding as input scaling increases, and demonstrates that combining training-free strategies like Audio-Permutational Self-Consistency with Chain-of-Thought can significantly improve performance.
Mousse is a novel optimizer that improves upon the Muon algorithm by integrating Shampoo's Kronecker-factored preconditioning to adaptively handle the heavy-tailed curvature of deep neural networks, thereby achieving faster training convergence with negligible computational overhead.
The paper introduces ActiveUltraFeedback, an efficient active learning pipeline that leverages uncertainty estimates and novel selection strategies like Double Reverse Thompson Sampling to generate high-quality preference data, enabling Large Language Models to achieve superior alignment performance with as little as one-sixth of the annotated data required by static baselines.
The paper proposes ESAinsTOD, a unified end-to-end schema-aware instruction-tuning framework that leverages full-parameter LLM fine-tuning with instruction and schema alignment mechanisms to achieve superior performance, generalization in low-resource settings, and robustness against noise across diverse task-oriented dialog benchmarks.
This study demonstrates that a custom Transformer architecture outperforms both traditional machine learning models and zero-shot generative LLMs in automatically classifying cardiac risk from large-context, unstructured Dutch electronic health records, offering a robust alternative to manual administrative coding for geriatric cardiovascular risk management.
This paper introduces the CoMoral benchmark to reveal that current Large Language Models exhibit a critical limitation where they prioritize moral reasoning over commonsense understanding, particularly struggling to detect contradictions involving primary characters due to a pervasive narrative focus bias.
This study investigates gender bias in Large Language Models within French healthcare contexts, demonstrating that these models rely on embedded stereotypes when processing interactions between gender and other social determinants of health, thereby highlighting the need for context-specific assessments that go beyond evaluating individual factors in isolation.
TaSR-RAG is a taxonomy-guided framework that enhances Retrieval-Augmented Generation for multi-hop reasoning by decomposing complex queries into structured triple sub-queries and performing step-wise evidence selection through hybrid matching, thereby achieving superior accuracy and clearer reasoning traces without relying on costly graph construction.
This paper investigates emotion as a latent factor influencing LLM attention and reasoning, introducing the AURA-QA dataset and an emotional regularization framework that demonstrably improves reading comprehension performance across both emotionally varying and standard benchmarks.
DuplexCascade is a VAD-free, cascaded ASR-LLM-TTS pipeline that enables full-duplex speech-to-speech dialogue by converting long turns into micro-turns and utilizing special control tokens to coordinate turn-taking while preserving the conversational intelligence of large language models.
The paper introduces BiCLIP, a simple and parameter-efficient framework that achieves state-of-the-art few-shot domain adaptation for vision-language models by applying a structured geometric transformation to align multimodal features across disparate domains using a small set of anchor samples.
The paper introduces VoxEmo, a comprehensive benchmark and toolkit for evaluating Speech Large Language Models on speech emotion recognition across 35 corpora and 15 languages, featuring a distribution-aware soft-label protocol that reveals how these models uniquely align with human subjective emotion distributions despite trailing supervised baselines in hard-label accuracy.
PathoScribe is a unified retrieval-augmented large language model framework that transforms static pathology archives into an active, reasoning-enabled clinical intelligence platform, enabling natural language case retrieval, automated cohort construction, and real-time diagnostic support with high accuracy and efficiency.
This paper introduces Fish Audio S2, an open-source text-to-speech system that leverages a multi-stage training pipeline to enable multi-speaker, multi-turn generation with natural-language instruction following, while providing production-ready weights and an efficient SGLang-based inference engine.
This paper introduces MathQ-Verify, a novel five-stage pipeline that rigorously filters ill-posed or under-specified mathematical questions through format validation, formalization, contradiction detection, and completeness checks, achieving state-of-the-art performance in curating reliable datasets for training large language models.