DataSTORM is an LLM‑based agentic system designed to conduct deep research over large‑scale structured databases and internet sources. It applies principles of Exploratory Data Analysis and Data Storytelling to frame research as a thesis‑driven analytical process, iteratively generating hypotheses, performing quantitative reasoning, and crafting coherent narratives. Evaluations on InsightBench and a new ACLED‑based dataset show that DataSTORM surpasses existing systems, achieving significant improvements in insight‑level recall and summary‑level scores.
By Shicheng Liu, Yucheng Jiang, Sajid Farook, Camila Nicollier Sanchez, David Fernando Castro Pena, Monica S. Lam
AEScorer is an agentic evidence‑grounded framework designed for graded factuality verification, addressing the limitation of binary judgments in current methods. It operates in two stages: first, it gathers and refines external evidence through agentic search; second, it predicts a scalar factuality score to capture nuanced differences in correctness. The authors also introduce GradedVeriBench, a benchmark covering general and multi‑hop question answering, and demonstrate that AEScorer outperforms existing approaches on this new benchmark.
By Hui Huang, Muyun Yang, Yuki Arase
The paper introduces STAR, a metric that measures sentence-level alignment between source and target documents in document-to-document machine translation. Using STAR, the authors develop StarPO, a preference‑optimization framework that ranks translation hypotheses by structural quality and applies a dynamic alignment mask to focus learning on misaligned segments. Experiments on news and literary data show that StarPO improves both translation quality and structural integrity, enabling small models to outperform large proprietary systems such as GPT‑4o while remaining more token‑efficient.
By Yichen Dong, Hao Wang, Junhui Li, Linlong Xu, Longyue Wang, Weihua Luo
The paper introduces Centroid Intervention Fusion (CIF), a framework that merges multiple multilingual intervention projections into a single language-shared operator for inference-time modification of large language models. CIF improves cross-lingual transfer without updating model parameters and achieves up to +3.378 percentage points better performance than prior pairwise intervention baselines across several benchmarks, including low-resource languages. The authors provide code at https://github.com/VRCMF/CIF.git.
By Wei Sun, Marie-Francine Moens
The paper surveys the state of Explainable AI (XAI) in Arabic NLP, highlighting three gaps: a method gap where Arabic XAI relies mainly on limited post‑hoc techniques; a task gap with most work focused on classification tasks and little on generation, retrieval, or dialogue; and a linguistic gap where explanations rarely address Arabic‑specific phenomena such as morphology, dialects, and diglossia. It proposes a taxonomy of tasks, methods, linguistic units, and evaluation practices, and outlines a research agenda for linguistically grounded Arabic XAI.
By Salima Lamsiyah, Ruslan Mitkov
The paper investigates whether a language model’s own confidence can replace labeled data for teaching it to abstain from uncertain answers. By fine‑tuning models with LoRA to answer only when their frozen confidence is high and to say “I’m not sure” otherwise, the authors show that this label‑free approach matches label‑supervised abstention tuning on short‑form factual QA. The method works across six open‑weight models (1B‑8B) and is effective except for confidently wrong facts, which the confidence signal cannot flag.
By Ali Asaria, Tony Salomone, Deep Gandhi
The paper shows that HuggingFace’s ByteLevel pre‑tokenizer, which treats a word as a sequence of Unicode letters, splits abugida scripts at every vowel sign, creating a training‑free lower bound on tokenizer fertility. Across 26 languages, all 17 abugidas exhibit increased token counts (up to 9×), while Latin, Cyrillic, Hangul, and Han remain unchanged. The authors demonstrate that correcting the character class reduces Nepali token counts, improves model performance, and that this issue is widespread in popular HuggingFace models.
By Sajal Regmi, Siddhartha Pudasaini, Chetan Phakami Pun
The paper investigates the linguistic characteristics of ChatGPT-generated text, comparing it to 1,000 scientific publications and exploring its relation to concepts of ‘bullshit’ in political speech and workplace contexts. By applying hypothesis‑testing methods, the authors demonstrate that a statistical model of bullshit can link the artificial bullshit produced by ChatGPT to the political and workplace functions of bullshit observed in natural human language.
By Alessandro Trevisan, Harry Giddens, Sarah Dillon, Alan F. Blackwell
The paper introduces MedREAL, a unified framework that aligns linguistic reasoning with spatial grounding for medical visual question answering and segmentation. MedREAL employs Seg Anchored Reasoning Pooling (SARP) to extract semantic evidence from segmentation tokens and a Reasoning-to-Visual (R2V) fusion mechanism to integrate these features into a segmentation pipeline. Using the newly created MedRAVS-13K dataset, MedREAL achieves superior performance, reporting 68.49% gIoU and 70.47% cIoU, and generates evidence masks that consistently match textual diagnoses.
By Haowen Gu, Gensheng Pei, Junzhu Mao, Qiong Wang, Mingwu Ren, Yazhou Yao
MedFG-VQA is a lightweight medical visual question answering framework that uses a memory bank to enhance low‑frequency DCT features and graph‑enhanced cross‑attention for visual‑textual alignment. It introduces Frequency‑Memory Fusion to retrieve and fuse low‑frequency information from a learnable memory bank, and Graph‑Aware Cross‑Attention to refine cross‑modal features via graph convolution. The authors also create SynMed‑VQA, a synthetic dataset of over 2 million QA pairs across nine imaging modalities, and show that MedFG‑VQA matches or outperforms larger models on several biomedical VQA benchmarks while keeping computational costs low.
By Haowen Gu, Gensheng Pei, Zeren Sun, Mingwu Ren, Xiangbo Shu, Yazhou Yao, Fumin Shen
The paper introduces an importance‑scoring metric for multi‑head transformer attention heads applied to tabular data, a domain where transformers have been less studied. Experiments on 40 diverse tabular datasets show that removing heads with the lowest importance scores has minimal impact on performance, while removing the most important head first causes the largest drop. The study finds that important heads are distributed across layers and vary significantly across different tabular schemas, suggesting that the proposed score can help reduce redundancy and improve transformer efficiency.
By Ahmad Jad Allah, Kazi F. Akhter, Md. Kamrozzaman Bhuiyan, Manar D. Samad
The paper investigates Self‑Generated Text Recognition (SGTR), the ability of large language models (LLMs) to identify their own outputs. By evaluating 13–21 models across 6 experimental designs, it shows that SGTR accuracy varies with evaluation format, conversation structure, and task domain, and that a quality‑heuristic bias dominates results. The study also finds that fine‑tuning for SGTR in one setting can generalize to others and may cause models to prefer their own outputs when judging, highlighting potential safety concerns.
By Jesse St. Amand, Callum Canavan, Sohaib Imran, Joseph Hewson, Aaron Lutz, Shi Feng, Puria Radmard, Lennie Wells
arXiv:2608. 27268v1 Announce Type: new Abstract: Although Large language models (LLMs) mediate access to knowledge and computational assistance, their capabilities should benefit vulnerable groups in the same way.
By Jinghan Zhang, Fengran Mo, Zhiyu Chen, Xiaoyan Han, Kunpeng Liu, Chang-Tien Lu
RuleWeaver is a benchmark construction framework designed to evaluate large language models’ ability to reason over complex, rule‑centered scenarios. It begins with corpus‑derived IF‑THEN meta rules, expands them into more intricate rules, and composes these into scenario‑based QA instances. The benchmark assesses not only final answer correctness but also process‑level metrics such as rubric‑based answer quality, rule recall, and rule precision, revealing that current LLMs achieve only about 50% of the maximum rubric score on these tasks.
By Bohan Yu, Shi-Yang Li, Pengfei Cao, Jun Zhao, Kang Liu
The paper introduces Self-Improving Retrieval-Augmented Generation (RAG), a framework that splits document question answering into Retrieval, Reasoning, and Judge agents coordinated by an orchestrator. When the Judge scores an answer below a dynamic threshold, the system retries with broader retrieval, more careful prompting, and relaxed acceptance criteria, achieving 86% oracle-guided accuracy on FinanceBench with a 36.4% Lazarus Rate. The approach logs every decision with confidence scores, providing audit trails needed for regulated financial applications.
By Junjie Xiong, Shawheen Ghezavat, Aum Hirpara
The paper introduces CAST, a concept-guided artifact suppression tuning framework that uses sparse autoencoders to identify and suppress note-specific artifacts in clinical language models. CAST labels latent features with an LLM-assisted pipeline and ICD‑10 constraints, then fine‑tunes the model while providing post‑hoc per‑concept attributions for auditability. In experiments on MIMIC‑IV discharge‑note mortality prediction, CAST outperforms standard fine‑tuned encoders and competes with strong LLM baselines while offering a feature‑level audit trail of clinical concepts and suppressed artifacts.
By Jin Mu, Guanhua Chen
CIFQA is a deterministic, tool‑grounded multi‑agent framework that separates language understanding from numerical execution for financial question answering. It assigns specialized agents for interpretation, routing, parameter extraction, computation planning, and response generation, while deterministic Python tools perform the calculations. On a fixed‑deposit benchmark, CIFQA achieves 95.54% accuracy on calculation‑intensive queries and 90.87% overall, outperforming larger LLM baselines and showing that architecture, not scale, drives numerical reliability.
By Kunjesh Parekh, Anil Kumar Tiwari, Divya Saxena
BanglaVerse is a new benchmark that evaluates multilingual vision‑language models on Bengali culture, covering nine visual domains and expanding to four languages and five Bangla dialects for a total of about 32,200 artifacts. It includes visual question answering and captioning tasks built from 1,152 manually curated images. Experiments show that models perform worse on dialectal variants and that missing cultural knowledge, rather than visual grounding, is the main bottleneck.
By Nurul Labib Sayeedi, Md. Faiyaz Abdullah Sayeedi, Shubhashis Roy Dipta, Mahbub E Sobhani, Rubaya Tabassum, Ariful Ekraj Hridoy, Mehraj Mahmood, Md. Tarek Hasan, Swakkhar Shatabda
The paper introduces MDiTFace, a diffusion transformer designed for high‑fidelity mask‑text collaborative facial generation. It unifies tokenization of semantic masks and text, enabling synchronous multimodal feature interaction via stacked multivariate transformer blocks. A novel decoupled attention mechanism separates dynamic and static computations, allowing caching of static features and reducing mask‑condition overhead by over 94% while preserving performance, leading to superior facial fidelity and conditional consistency compared to existing methods.
By Yushe Cao, Dianxi Shi, Xing Fu, Xuechao Zou, Haikuo Peng, Xueqi Li, Chun Yu, Junliang Xing
Cascaded Batch Prompting introduces a two‑stage method that separates complex reasoning from symbol grounding to address the unpredictability of conventional batch prompting. Experiments on multiple‑choice question answering and natural language inference show that this approach outperforms standard single prompting while maintaining a speedup proportional to batch size. The technique establishes a new state‑of‑the‑art position on the Pareto frontier for efficiency and performance.
By Sho Hoshino, Peinan Zhang