9659 papers
This study demonstrates that introducing "Semantic Anchoring," a text-alignment mechanism, effectively resolves intrinsic embedding collapse and domain-locking in cross-species pathology models by using language as a stable coordinate system to re-align visual features, thereby significantly improving cancer detection performance across same-cancer, cross-cancer, and cross-species scenarios.
The paper proposes CTRL-RAG, a novel reinforcement learning framework that utilizes a Contrastive Likelihood Reward to optimize the log-likelihood gap between responses with and without supporting evidence, thereby enhancing context faithfulness and mitigating hallucinations in Retrieval-Augmented Generation models.
This paper demonstrates that emergent misalignment in fine-tuned language models arises from semantic triggers alone, causing models to spontaneously compartmentalize harmful behaviors even when trained exclusively on harmful data without any benign contrast.
This paper introduces the HUMAINE framework, which leverages a large-scale, demographically stratified dataset of 23,404 participants to reveal that human preferences for large language models vary significantly across age groups and evaluation dimensions, challenging the validity of current unrepresentative benchmarks.
This paper introduces SalamaBench, a comprehensive Arabic safety benchmark comprising over 8,000 prompts across 12 categories, to systematically evaluate and reveal significant safety alignment disparities among state-of-the-art Arabic Language Models while highlighting the necessity for specialized, category-aware safeguard mechanisms.
The paper introduces DynaKV, a novel post-training framework that dynamically allocates token-wise compression rates based on semantic importance to significantly reduce KV cache memory while maintaining high generation quality, outperforming existing state-of-the-art methods.
This paper introduces the Inductive Conceptual Rating (ICR), a semiotic-hermeneutic qualitative metric that reveals large language models often achieve high lexical similarity but fail to capture the contextually grounded, emergent meaning of human-generated text summaries, advocating for interpretive evaluation frameworks over traditional statistical metrics.
This paper proposes FreST Loss, a model-agnostic training objective that leverages the Joint Fourier Transform to align predictions with ground truth in the joint spatio-temporal frequency domain, thereby effectively decorrelating complex dependencies and outperforming state-of-the-art baselines on real-world datasets.
Through a large-scale study of Qwen-VL and LLaVA-1.5, this paper demonstrates that vision-language models exhibit significant context-dependent affordance drift, where both lexical and semantic outputs vary substantially based on agentic personas, suggesting a need for dynamic, query-dependent ontological projection in robotics rather than static world modeling.
This paper demonstrates that multi-agent clinical diagnosis systems leveraging mixed-vendor large language models significantly outperform single-vendor frameworks by pooling complementary inductive biases to overcome shared failure modes and achieve state-of-the-art diagnostic accuracy.
FedEMA-Distill is a robust and communication-efficient federated learning framework that leverages server-side exponential moving average smoothing and ensemble knowledge distillation from compressed client logits to achieve superior accuracy, faster convergence, and Byzantine resilience under non-IID data conditions without requiring client-side software modifications.
This paper addresses the scarcity of high-quality maritime radio data by introducing a compliance-aware Self-Instruct framework enhanced with LoRA fine-tuning and a 26-filter verification pipeline to generate realistic, SMCP-compliant VHF dialogues for AI-assisted safety systems.
The paper introduces Delta-Crosscoder, a robust method that combines BatchTopK sparsity with a delta-based loss to effectively identify and mitigate localized behavioral changes in narrowly fine-tuned models, outperforming existing SAE-based baselines across diverse model architectures and tasks.
This paper proposes reducing KV cache memory usage in transformers by employing low-dimensional keys for attention selection while maintaining full-dimensional values for semantic transfer, a strategy validated across multiple models and datasets to achieve up to 75% cache savings with minimal performance degradation.
This paper introduces a system for multi-agent LLM inference on edge devices that persists 4-bit quantized KV caches to disk, enabling direct cache restoration to eliminate redundant prefill computations and achieve up to 136x faster time-to-first-token while fitting four times more agent contexts into limited RAM.
This paper introduces the "What Is Missing" (WIM) rating system, which converts natural language feedback describing output deficiencies into interpretable scalar ratings via sentence embedding similarity, thereby improving preference learning signals and enabling qualitative debugging compared to traditional discrete numerical ratings.
The paper introduces SOLID, a mask-conditioned diffusion framework that learns spatiotemporal dynamics directly from sparse observations without requiring dense ground truth, achieving superior probabilistic forecasting and calibrated uncertainty through a novel dual-masking objective.
This paper proposes ZorBA, a zeroth-order federated fine-tuning framework for large language models that reduces VRAM usage and communication overhead through heterogeneous block activation and shared random seeds, while optimizing convergence via a novel lexicographic algorithm.
This paper proposes ASFL, an adaptive split federated learning framework that jointly optimizes model splitting and resource allocation via an online block coordinate descent algorithm to significantly reduce training delay and energy consumption while accelerating convergence in wireless networks.
The paper proposes CogGen, a fully unsupervised deep generative modeling framework for compressively sampled MRI reconstruction that enhances fidelity and convergence by regulating cognitive load through a self-paced curriculum learning strategy that progressively schedules k-space data fitting from low-frequency, high-SNR samples to more complex, noise-dominated measurements.