1067 papers
This paper proposes a dynamic knowledge fusion framework for multi-domain dialogue state tracking that addresses challenges in modeling dialogue history and data scarcity by using a contrastive learning-based encoder to select relevant slots and leveraging their structured information as contextual prompts to improve tracking accuracy and generalization.
This paper systematically analyzes widely used speech tokenizers and reveals that they primarily encode phonetic rather than lexical-semantic information, highlighting a critical mismatch with text-derived semantics that necessitates new design approaches for effective multimodal LLM integration.
This paper introduces SearchLLM, the first large language model designed for open-ended generative search on platforms like RedNote, which utilizes a hierarchical multi-dimensional reward system and Gated Aggregation Strategy with GRPO to balance safety, factual grounding, and user alignment, resulting in measurable improvements in generation quality and user engagement.
This paper introduces a multi-agent negotiation framework that trains large language models to align with Collective Agency and resolve value conflicts through self-play deliberation optimized via RLAIF with GRPO, demonstrating improved conflict-resolution capabilities without compromising general language performance.
This paper introduces PEEM, a unified framework that employs a structured 9-axis rubric and LLM-based evaluators to provide interpretable, joint assessments of prompts and responses, enabling systematic diagnosis and significantly improving downstream accuracy through zero-shot prompt optimization.
The paper introduces PULSE, a medical reasoning agent that integrates a domain-tuned large language model with scientific literature retrieval to achieve expert-competitive diagnostic accuracy across varying disease incidences, while demonstrating both its potential to enhance physician decision-making and the risks of automation bias in collaborative workflows.
The paper introduces VERI-DPO, an evidence-aware alignment framework that leverages claim verification to mine preference pairs for Direct Preference Optimization, significantly reducing unsupported claims and improving the faithfulness of clinical summarizations while maintaining informative length.
This paper introduces VeriEnv, a framework that leverages language models to automatically clone real-world websites into safe, verifiable synthetic environments, enabling autonomous web agents to perform scalable self-evolution and achieve strong generalization without relying on unsafe real-world interactions or heuristic reward signals.
The paper introduces IH-Challenge, a reinforcement learning dataset designed to enhance instruction hierarchy robustness in frontier LLMs, which significantly improves their ability to prioritize instructions against conflicts and adversarial attacks while maintaining helpfulness and minimizing capability regression.
The AILS-NTUA system achieves top rankings in SemEval-2026 Task 8 by employing a unified architecture that prioritizes query diversity over retriever diversity for passage retrieval and utilizes a multistage generation pipeline with answerability calibration to optimize response quality.
This paper introduces Group Relative Reward Rescaling (GR), a novel reinforcement learning method that effectively mitigates length inflation in large language models by reframing length control as a multiplicative rescaling paradigm, thereby achieving lossless optimization and superior performance compared to existing baselines without compromising downstream capabilities.
This paper introduces an automatic end-to-end data integration pipeline powered by GPT-5.2 that generates all necessary configuration artifacts, demonstrating comparable or superior performance to human-designed pipelines across multiple case studies at a significantly lower cost.
This paper proposes a scalable, end-to-end automatic evaluation framework that generates Q&A pairs from knowledge bases, leverages LLMs for response assessment, and employs uncertainty filtering to significantly reduce human review costs while maintaining high agreement with manual judgments.
This paper empirically demonstrates that contrary to the hypothesis that moral reasoning alignment requires diversity-seeking algorithms, standard reward-maximizing RLVR methods are equally or more effective because high-reward moral responses exhibit a concentrated distribution in semantic space similar to logical reasoning tasks.
The paper introduces MUNIChus, the first multilingual benchmark for news image captioning comprising nine languages including low-resource ones, to address the scarcity of non-English datasets and evaluate the performance of state-of-the-art models in this challenging task.
This paper introduces a neural scoring function trained on an LLM-annotated synthetic benchmark to disentangle taxonomic similarity and thematic relatedness in topic models, demonstrating that these distinct semantic dimensions not only characterize differences across model families but also predict downstream task performance.
This paper proposes Conditional Expectation Reward (CER), a novel reinforcement learning method that utilizes the large language model itself as an implicit verifier to provide soft, graded reward signals, thereby overcoming the limitations of rule-based verification and enabling effective reasoning training across both mathematical and general free-form answer domains.
This paper outlines a research program focused on evaluating and enhancing the reasoning capabilities of the Polish large language model Bielik through benchmarking, comparative analysis, and strategic planning to ensure its competitiveness in the evolving AI landscape.
The paper introduces DxEvolve, a self-evolving diagnostic agent that emulates clinician cognition through an interactive deep clinical research workflow, autonomously requisitioning examinations and externalizing experience to achieve superior diagnostic accuracy and governed continual improvement compared to existing AI models.
This paper introduces EvoSchema, a comprehensive benchmark featuring a novel taxonomy of ten schema perturbation types to evaluate and enhance the robustness of text-to-SQL models against real-world database schema evolution, revealing that table-level changes significantly impact performance and demonstrating that training on diverse schema designs improves model resilience.