1077 papers
The paper introduces POET-X, a memory-efficient and scalable variant of the POET framework that utilizes optimized orthogonal equivalence transformations to enable the stable pretraining of billion-parameter large language models on a single GPU, overcoming the high memory and computational costs of the original implementation.
This paper proposes a novel diffusion-based method for Open-Vocabulary Camouflaged Instance Segmentation (OVCIS) that effectively fuses multi-scale textual-visual features to overcome the challenges of blending boundaries and segmenting unseen object classes, demonstrating superior performance on benchmarks with applications in surveillance, wildlife monitoring, and military reconnaissance.
This paper proposes RAEE, a robust retrieval-augmented framework that models early exit as a distribution prediction problem to leverage similar data's exit information for guiding adaptive layer termination, thereby accelerating LLM inference while maintaining or enhancing zero-shot performance across multiple tasks.
This paper proposes a novel framework that integrates Large Language Models with the Australian National University's Computer Science Scholarly Knowledge Graph, utilizing a Deep Document Model and KG-enhanced Query Processing to enable accurate, fine-grained semantic retrieval of research artifacts through automatic LLM-SPARQL fusion.
This study demonstrates that neural network language models can learn English passivization exceptions by leveraging both frequency-based entrenchment and semantic affectedness from their training data, validating the utility of manipulating training corpora to investigate the sources of linguistic evidence in acquisition.
This paper introduces LMUnit, a framework that utilizes natural language unit tests and a unified scoring model to overcome the limitations of human and automated evaluation, significantly improving inter-annotator agreement and achieving state-of-the-art performance on key benchmarks.
This paper identifies and empirically validates "preference leakage," a pervasive contamination problem where LLM judges exhibit significant bias toward student models with which they share a relationship (such as being the same model, having an inheritance link, or belonging to the same family), thereby challenging the reliability of LLM-as-a-judge evaluation paradigms.
OSCAR is a novel online soft compression and reranking method that dynamically reduces computational overhead in Retrieval-Augmented Generation pipelines at inference time, achieving significant speed-ups with minimal accuracy loss across various large language model sizes.
This paper introduces \data, a dataset of fine-grained interaction prompts, and proposes \model, a novel framework leveraging Multimodal Large Language Models for prompt decomposition, image critique, and targeted refinement to significantly enhance the generation of complex object interactions in text-to-image synthesis.
This paper introduces Noise-to-Meaning Recursive Self-Improvement (N2M-RSI), a minimal formal model demonstrating that AI agents feeding their own outputs back as inputs can trigger unbounded internal complexity growth once a specific information-integration threshold is crossed, while remaining implementation-agnostic and scalable to agent swarms.
This paper introduces PubHealthBench, a new benchmark comprising over 8,000 questions derived from UK government guidance, to evaluate LLMs on public health knowledge and finds that while state-of-the-art models excel in multiple-choice tasks, their performance on free-form responses remains limited, highlighting the need for additional safeguards in real-world applications.
This paper introduces Multi-Objective Balanced Covering (MoB), a novel visual token pruning framework that leverages Hausdorff distance and -covering theory to derive a closed-form error bound and dynamically balance prompt alignment with visual preservation, achieving significant inference acceleration with minimal performance loss across diverse multimodal models.
The paper introduces R1-Code-Interpreter, a general-purpose LLM trained via multi-stage reinforcement learning and supervised fine-tuning on a diverse dataset of 144 tasks, which significantly improves reasoning accuracy and enables emergent self-checking behaviors, outperforming both text-only and tool-augmented GPT-4o models.
This paper introduces Supervised Calibration (SC), a loss-minimization framework that enhances In-Context Learning in Large Language Models by learning optimal per-class affine transformations to correct systematic biases and alter decision boundary orientations, thereby achieving state-of-the-art performance across multiple models and datasets.
This paper investigates how preference models overrely on superficial artifacts like length and style rather than substantive quality, quantifying this miscalibration against human preferences and demonstrating that a counterfactual data augmentation method effectively mitigates these biases while preserving overall performance.
This paper introduces CounselBench, a large-scale benchmark developed with mental health professionals to evaluate and adversarially stress-test large language models on realistic open-ended mental health questions, revealing significant gaps in safety, personalization, and the reliability of automated evaluation compared to human expert judgment.
This paper proposes "Internal Confidence," a novel, training-free method that estimates query-level uncertainty by leveraging self-evaluations across model layers and tokens to detect knowledge boundaries, thereby enabling more efficient and trustworthy adaptive inference strategies like retrieval-augmented generation and model cascading.
This paper proposes a novel context biasing method for automatic speech recognition that leverages user-provided on-the-fly corrections of substitution errors to effectively resolve pronunciation-orthography mismatches, achieving a 22% to 34% relative improvement in biased word error rates without compromising overall system performance.
The paper introduces RLVER, a novel reinforcement learning framework that utilizes verifiable emotion rewards from simulated users to significantly enhance the empathetic capabilities of large language models while preserving their logical and coding competencies.
The paper introduces UQLM, a Python package that leverages state-of-the-art uncertainty quantification techniques to generate confidence scores for detecting hallucinations and enhancing the reliability of Large Language Model outputs.