arXiv:2602. 02498v2 Announce Type: replace-cross Abstract: Large language models can produce toxic or inappropriate text even for benign inputs, creating risks when deployed at scale.
By Baturay Saglam, Dionysis Kalogerias
The paper introduces Transition-Aware Reward-Guided Representation Sampling (TARGRS), a test-time detoxification framework for large language models that learns reward guidance over toxicity transitions in latent representation space. TARGRS constructs trajectories from non‑toxic to toxic states, uses toxicity scores along these paths to train a token‑level reward model, and then guides coarse‑to‑fine sampling in the opposite direction to produce detoxified outputs with minimal perturbation. Experiments on ten LLMs and three LVLMs show that TARGRS consistently outperforms existing baselines, reducing toxicity by up to 69.38% on RealToxicityPrompts and up to 64.23% on LVLMs while maintaining near‑original speed and core capabilities.
By Aishan Liu, Yisong Xiao, Xianglong Liu, Yongxin Huang, Siyuan Liang, Zonghao Ying, Yang Qu, Jian Yang, Yaodong Yang, Dacheng Tao
The Writerslogic team participated in the CLEF 2026 SimpleText shared task, tackling both text simplification (Task 1) and complexity spotting (Task 2). For simplification, they built a multi‑candidate pipeline with GPT‑4o‑mini, selecting the best candidate via a reference‑free heuristic, and their Claude Sonnet 4 submission achieved a SARI of 47.43 and BLEU of 14.21, ranking third overall on the Task 1 leaderboard. For complexity spotting, they fine‑tuned a DeBERTa‑v3‑large NLI model on 350 K labeled pairs, achieving a macro F1 of 0.8081 (0.8085 in an ensemble) on binary over‑generation identification and 0.804 accuracy on multi‑class error classification, placing them second among unique teams.
By David L. Condrey
arXiv:2609.39027v1 Announce Type: new
Abstract: AI reviewers can assign different judgments to manuscripts that report the same science in different wording, potentially rewarding rhetorical optimiza...
By Chenguang Wang, Ming Li, Chengrui Fan, Jianpeng Chen, Han Chen, Tianyi Zhou, Dawei Zhou
DeepSafe‑SciHazard introduces a real‑world‑grounded benchmark for measuring scientific safety risks, comprising 2,400 hazardous and 600 oversafety questions across 12 disciplines. It defines a decomposed harm scoring system, DeHarm‑Score, that evaluates query severity, refusal behavior, and response‑level risk, further breaking down harm into executability and net‑new risk. An expert‑validation study shows DeHarm‑Score aligns 90.17% better with expert judgments than the best baseline, and the benchmark reveals that deep research agents score 32.3% higher on mean DeHarm‑Score than standard LLMs, highlighting a critical blind spot in current safety defenses.
By Chunxiao Li, Yuan Xiong, Lijun Li, Tianyi Du, Wenlong Zhang, Lei Bai, Jing Shao
The paper introduces an LLM-as-a-Judge framework for evaluating the outputs of an agentic drug discovery assistant, ChatInvent, deployed at AstraZeneca. It defines four quality dimensions—Completeness, Relevancy, Structural Clarity, and Scope Adherence—alongside deterministic Tool Call Correctness checks, and validates the judge against five expert annotators. After optimizing the best-performing judge with few-shot demonstrations, alignment with human majority votes improves from 0.80 to 0.86, and the framework reveals that informal question phrasing does not degrade output quality.
By Emma Granqvist, Roc\'io Mercado, Samuel Genheden