The paper introduces REASONS, a benchmark of 12,723 sentence-level citation instances across 12 arXiv subject categories, to evaluate scientific citation attribution under different evidence conditions. It proposes a dual-metric framework—Abstention Rate (AR) and Hallucination Rate (HR)—to balance reliability and responsiveness. Experiments with proprietary and open-source LLMs across various prompting and retrieval settings show that advanced Retrieval-Augmented Generation (RAG) reduces hallucinations but increases abstention, while adversarial metadata can push hallucination rates above 85%. Human evaluation confirms a high ratio of factual hallucinations to acceptable paraphrases, underscoring the need for systems that can appropriately abstain under uncertainty.
By Deepa Tilwani, Yash Saxena, Seyedali Mohammadi, Ankur Padia, Edward Raff, Amit Sheth, Srinivasan Parthasarathy, Manas Gaur
arXiv:2609.08709v1 Announce Type: new
Abstract: Designing foundation models for graphs is challenging due to the irregular structure of graphs and the different sizes and characteristics of embedding...
By Jonathan Frank, David Richerby, Ansgar Scherp
arXiv:2609.07050v1 Announce Type: new
Abstract: Retrieval-augmented generation (RAG) critically depends on retrieving the evidence necessary for effective reasoning. However, this remains particularl...
By JungMin Yun, YoungBin Kim
The study demonstrates that contrastive pre‑training of ECG representations using cardiac magnetic resonance (CMR) imaging data can enhance ECG‑based detection of Chagas disease. By aligning an ECG encoder with a CMR embedding space from 63,193 paired UK Biobank examinations, the authors achieved higher AUROC and sensitivity metrics on CODE‑15%, SaMi‑Trop, and PhysioNet/CinC 2025 Challenge datasets compared to an unaligned baseline. The approach shows that imaging‑supervised ECG representations generalize across different populations and resource‑constrained settings.
By Laura Alvarez-Florez, Daniel Uyterlinde, Samuel Ruip\'erez-Campillo, Lukas P. A. Arts, Folkert W. Asselbergs, Fleur V. Y. Tjong
arXiv:2609.05919v1 Announce Type: new
Abstract: We propose a graph dictionary learning (GDL) framework where each graph is represented as a zero-mean Gaussian distribution derived from its filtered L...
By Jinchuan Liao, Dai Hai Nguyen
arXiv:2609.08899v2 Announce Type: replace-cross
Abstract: Speech deepfakes can mimic a speaker's voice convincingly enough to deceive listeners and automated systems. This has driven strong progress...
By Mengzhe Geng, Yujia Lu, Patrick Littell, Manuela Kunz, Xie Chen
arXiv:2609.05518v1 Announce Type: cross
Abstract: Despite the strong capabilities of multimodal large language models (MLLMs), their parametric knowledge remains incomplete and difficult to update, m...
By Jiacheng Cai, Zijin Hong, Zheng Yuan, Huachi Zhou, Qinggang Zhang, Xiao Huang
The paper explores how individual text corpora (ITCs) can be integrated into small language models (SLMs) using DoRA fine‑tuning. By training a DoRA adapter for each of 150 participants, the authors show that the adapter can encode a participant’s own ITC into the model’s weights, improving fit to that participant’s held‑out text. However, while the adapter improves log‑loss on a generalized knowledge test, it does not enhance accuracy under a bias‑corrected PMI readout, and adding retrieval does not provide further benefit.
By Christoph Wigbels, Ali Abusaleh, Markus T. Jansen, Alexander Mehler, Markus J. Hofmann
The paper investigates how many transformer components influence a token prediction by measuring the absolute contribution of each unit and channel to the logit. It finds that thousands of components contribute to a single prediction, yet a small subset—often just dozens—carries the majority of the predictive mass. Across models ranging from 124 M to 7 B parameters, the proportion of the model involved in a prediction remains around one to three percent, independent of size, and the study demonstrates that specific components can be directly read and written to modify model behavior without additional training.
By Mark Oskin
EvidenceNet is a disease‑specific dataset that transforms full‑text biomedical literature into structured evidence records and graph representations, preserving study design, provenance, and quantitative support. Using an LLM‑assisted pipeline, it extracts experimentally grounded findings, normalizes entities, scores evidence quality, and links related records via typed semantic relations. The released subsets—EvidenceNet‑HCC and EvidenceNet‑CRC—contain thousands of evidence records and richly connected graphs, with high extraction and relation‑type accuracy, enabling retrieval‑augmented question answering and graph‑based tasks such as link prediction and target prioritization.
By Chang Zong, Jinyu Chen, Sicheng Lv, Si-tu Xue, Huilin Zheng, Jian Wan, Lei Zhang
arXiv:2605.27156v2 Announce Type: replace-cross
Abstract: Retrieval-Augmented Generation (RAG) enhances Large Language Models (LLMs) by incorporating external knowledge, particularly for long-tail do...
By Ruikang Zhang, Zhanni Chen, Yiqiao Cai, Qi Su
The paper investigates how different components of a graph retrieval‑augmented generation pipeline affect large language model performance on knowledge‑graph question answering. It examines four variables—whether the answer path is included, the syntax of triples, the order of triples, and the subgraph size—across six LLMs and two benchmarks. The study finds that including the answer path is crucial, while the grounding instruction dramatically reduces accuracy when no facts are provided, and that syntax, order, and subgraph size have negligible measurable impact at multi‑hop depth.
By Arquimedes Canedo
RubricRefine is a training‑free pre‑execution refinement method that generates task‑specific rubrics from tool documentation, scores candidate code against explicit contract checks, and iteratively repairs failures before execution. It achieves an average score of 0.86 across seven models on M3ToolEval without any execution attempts, outperforming prior inference‑time baselines while incurring lower latency. The approach shows consistent performance on single‑step API‑Bank tasks and maintains an advantage in multi‑turn settings on AppWorld, with its effectiveness tied to the quality of the supplied documentation.
By Will LeVine, Brendan Evers, Sam Saltwick, Abhay Venkatesh
arXiv:2512.03491v3 Announce Type: replace
Abstract: Neural Networks are indispensable to natural sciences and society. Their impact extends from applications in public health to workforce productivit...
By Antonin Sulc, Noor Naddour
The paper introduces MERIT, a benchmark that evaluates the marginal benefit of long‑term memory for tool‑using large language model agents while explicitly accounting for cost. MERIT provides episodic tool‑use tasks across three domains, verifies dependence on earlier‑episode facts, and measures memory operations in tokens and dollars. Experiments on GPT‑4.1‑mini, Claude Haiku 4.5, and Claude Sonnet 5 show that memory can significantly improve task success, but its utility varies widely across models and memory implementations, and full replay is rarely cost‑effective.
By Shweta Mishra, Shashank Mishra
The paper presents three Isabelle/HOL embeddings of monadic second‑order logic (MSO): a deep embedding, a maximal‑shallow embedding, and a minimal‑shallow embedding that collapses formulas to bool. It introduces a two‑sorted substitution system that ensures capture‑avoiding substitution and proves the faithfulness of all embeddings. A fully mechanised two‑sorted downward Löwenheim‑Skolem theorem is established, showing that the minimal embedding recovers deep validity relative to countable assignments and aligns with both the general (Henkin‑style) and standard readings of MSO, while also demonstrating differences in classical MSO properties across the embeddings.
By Christoph Benzmueller, Daniel Kirchner
The paper introduces a multilingual, multi-functional framework for disambiguating funder names in scientific publications, using a training dataset that merges the Research Organization Registry with Web of Science and Crossref Open Funder Registry data. By applying multi-task learning with contrastive and multiple negatives ranking losses, the authors fine‑tune open‑weight embedding models from the Sentence Transformer, Gemma, and Qwen3 families, achieving over 90% accuracy in matching Web of Science funder names to ROR identifiers and surpassing general‑purpose LLMs by more than 0.1. For funders not present in ROR, a similarity network is constructed to identify clusters, and the study discusses challenges related to smaller and non‑English‑speaking funders.
By Kanyao Han, Zhiwen You, Jinseok Kim, Jana Diesner
The paper introduces Evidence-Aligned Entity Verification (EAEV), a method for detecting entity-level hallucinations in retrieval-augmented generation (RAG). EAEV aligns generated entities with retrieved evidence across three dimensions and uses counterfactual stability analysis to maintain robust alignments when evidence changes. Experiments on multiple RAG benchmarks show that EAEV consistently outperforms existing hallucination detection methods and generalizes well.
By Runsong Jia, Zhen Fang, Mengjia Wu, Jie Lu, Yi Zhang
arXiv:2509.12635v4 Announce Type: replace-cross
Abstract: We prove under practical assumptions that Rotary Positional Embedding (RoPE) introduces an intrinsic distance-dependent bias in attention sco...
By Yu Wang, Sheng Shen, R\'emi Munos, Hongyuan Zhan, Yuandong Tian
arXiv:2603.01690v3 Announce Type: replace-cross
Abstract: While dense biomedical embeddings achieve strong performance, their opaque dimensions limit transparency in biomedical NLP. Recent question-b...
By Yixuan Tang, Zhenghong Lin, Yandong Sun, Wynne Hsu, Mong Li Lee, Anthony K. H. Tung