1071 papers
The paper introduces DataChef-32B, a reinforcement learning-based system that automates the end-to-end generation of optimal data recipes for adapting Large Language Models to specific tasks, achieving performance comparable to or exceeding human-curated pipelines and official checkpoints.
This systematic literature review critiques the "ground truth" paradigm in machine learning as a positivistic fallacy that misinterprets human disagreement as noise, arguing instead for pluralistic annotation infrastructures that treat diverse subjective perspectives as high-fidelity signals essential for building culturally competent models.
This paper introduces IntelliAsk, a question-generation model trained via RLVR with a novel reward model (IntelliReward) and DAPO optimization to produce high-quality, evidence-based research questions that outperform human reviewers and strong baselines in expert evaluations while also enhancing broader reasoning and writing capabilities.
This paper proposes a revised annotation scheme for cross-document coreference resolution that treats coreference chains as discourse elements to better capture lexical diversity and framing variations in news media, demonstrating through the reannotation of NewsWCL50 and ECB+ datasets that this approach enables more balanced and discourse-aware analysis.
This paper challenges the assumption that Speech LLMs inherently outperform ASRLLM pipelines by demonstrating through matched-backbone testing and mechanistic analysis that current Speech LLMs often function as expensive cascades relying on text representations, which can even underperform traditional pipelines under noisy conditions.
This paper introduces novel "Text-to-Big SQL" evaluation metrics to address the limitations of existing benchmarks in assessing production-level LLM agents, demonstrating that traditional Text-to-SQL metrics fail to capture critical cost, latency, and efficiency implications that arise when scaling to large datasets.
This paper reframes the modality collapse observed in multimodal LLMs as a mismatched decoding problem, demonstrating through information-theoretic analysis and empirical validation that the accessibility of non-text information is fundamentally limited by the decoder's training objective and scoring rule rather than the encoder's architecture or alignment.
The paper proposes CoME, a novel mobile agent architecture that employs four specialized experts with a progressive training strategy and an InfoGain-Driven DPO method to achieve balanced, decoupled enhancement of hybrid reasoning capabilities, outperforming existing dense and MoE approaches on AITZ and AMEX datasets.
This paper introduces Aletheia, a novel LLM-powered browser extension that combines Retrieval-Augmented Generation with interactive user features to effectively detect fake news and provide transparent, evidence-based explanations, outperforming existing baselines in both detection accuracy and user engagement.
The paper introduces Omni-C, a single dense Transformer encoder that compresses heterogeneous modalities (text, audio, and image) into shared representations via unimodal contrastive pretraining, thereby eliminating the parameter overhead and routing complexity of Mixture-of-Expert architectures while achieving comparable performance with significantly reduced memory usage.
This paper establishes that while language-equivalent context-free grammars yield identical token masks in grammar-constrained decoding, their structural differences significantly impact computational efficiency by introducing variable state-space blowups and ambiguity costs, leading to fundamental lower bounds on decoding work and new distortion metrics for masked sampling.
The paper introduces EigenData, a self-evolving multi-agent platform that automates the synthesis, auditing, and repair of high-quality function-calling training data, demonstrating its effectiveness by systematically correcting the Berkeley Function-Calling Leaderboard (BFCL-V3) to achieve model rankings that better correlate with human judgments of functional correctness.
This paper proposes CDDS, a novel cross-modal alignment algorithm that utilizes a dual-path UNet for constrained decoupling of semantic and modality components and a distribution sampling method to bridge the modality gap, thereby achieving superior semantic consistency and outperforming state-of-the-art methods by 6.6% to 14.2%.
The paper introduces CBR-to-SQL, a Case-Based Reasoning framework that improves Text-to-SQL generation in healthcare by utilizing abstract case templates and a two-stage retrieval process to achieve higher accuracy, sample efficiency, and robustness compared to standard Retrieval-Augmented Generation methods on the MIMICSQL dataset.
The paper introduces NOTAI.AI, an explainable framework that combines curvature-based signals, neural, and stylometric features within an XGBoost classifier to detect machine-generated text while using SHAP and an LLM layer to generate structured, natural-language rationales for its decisions.
This paper investigates how Chain-of-Thought prompting exacerbates the leakage of personally identifiable information in large language models, demonstrating that leakage varies significantly by model family and reasoning budget, and evaluating various lightweight inference-time gatekeepers to propose hybrid policies that balance reasoning utility with privacy protection.
The paper introduces RACAS, a robot-agnostic agentic system that uses natural language communication between LLM/VLM-based modules to control diverse robotic platforms without requiring code modifications or retraining, successfully demonstrating its effectiveness across wheeled, multi-jointed, and underwater robots.
This study demonstrates that large language models' moral judgments are highly fragile and manipulable, as they are significantly more influenced by narrative perspective, persuasion cues, and evaluation protocols than by the underlying moral substance of a dilemma.
FreeTxt-Vi is a free, open-source web toolkit that integrates a benchmarked bilingual Vietnamese-English NLP pipeline for segmentation, sentiment analysis, and summarization, enabling non-programmers to analyze text data while demonstrating competitive performance against existing baselines.
The paper introduces Ptychi-Evolve, an autonomous framework that leverages large language models and evolutionary mechanisms to automatically discover and evolve novel regularization algorithms for ptychography, achieving significant reconstruction quality improvements across diverse imaging datasets.