Yifei Yuan, Luis Salamanca, Sophia Schlosser, Laurence Brandenberger · May 6, 2026 · Citations: 0
Pairwise PreferenceGeneral
Approximating the ideological position of Members of Parliament (MPs) is a fundamental task in political science, helping researchers understand legislative behavior, party alignment, and policy preferences.
Zhipeng Song, Yizhi Zhou, Xiangyu Kong, Jiulong Jiao, Xuezhou Ye, Chunqi Gao · May 6, 2026 · Citations: 0
Pairwise PreferenceAutomatic MetricsGeneral
CAR converts these confidence changes into coarse precedence constraints and returns the feasible ranking with minimum Kendall distance from the baseline, preserving existing pairwise preferences unless generator-side evidence supports…
In the BM25-centered main analysis, CAR achieves a +5.53\% mean relative NDCG@5 gain; on the fixed NQ-answerable downstream evaluation, it improves token-level F1 by +0.43 points, with ranking and generation gains strongly aligned across…
Ziping Ye, Gourab Dey, Christos Christodoulopoulos, Charith Peris, Anil Ramakrishna, Weitong Ruan · May 5, 2026 · Citations: 0
Pairwise PreferenceAutomatic MetricsGeneral
In contrast to existing watermarking methods, which typically encode signatures by adjusting token selection preferences during text generation, SWAN embeds the signature directly in the sentence's semantic representation.
Empirical evaluation on the RealNews benchmark shows SWAN matches state-of-the-art detection performance on unaltered watermarked text, while significantly improving robustness against paraphrasing, increasing detection AUC by up to 13.9…
We evaluate on HotpotQA-RAG v3, a controlled multi-hop benchmark, under an artifact-aware protocol (shortcut baselines, counterfactual swaps, no-oracle checks, GPT-4o audits).
Calibrated SURE-RAG reaches 0.9075 Macro-F1 (0.8951 +/- 0.0069), substantially above DeBERTa mean-pooling (0.6516) and a GPT-4o judge (0.7284), while matching a strong but opaque concat cross-encoder (0.8888 +/- 0.0109) with full…
As the ecosystem of Large Language Model (LLM)-based agents expands rapidly, efficient and accurate Agent Discovery becomes a critical bottleneck for large-scale multi-agent collaboration.
To bridge this gap, we propose GRAIL (Granular Resonance-based Agent/AI Link), a novel framework achieving sub-400ms discovery latency without compromising accuracy.
Sadia Asif, Mohammad Mohammadi Amiri · May 3, 2026 · Citations: 0
Red TeamGeneral
Motivated by these findings, we introduce REFUSALGUARD, a representation-level fine-tuning framework that preserves safety-relevant structure during model adaptation.
We evaluate REFUSALGUARD across multiple model families, including LLaMA, Gemma, and Qwen, on adversarial safety benchmarks such as AdvBench, DirectHarm4, and JailbreakBench, as well as downstream utility tasks.
Multi-agent debate (MAD), and more broadly closed-system reasoning where agents iteratively transform each other's outputs, tends to preserve answer accuracy while degrading the reasoning behind those answers.
An R6 cohort study (Korean n=10x30 FEVER; English n=3x200 SciFact) finds inter-rater Fleiss kappa <= +0.018 with 0.8-1.4 Likert intra-rater shifts across language and domain -- the human agreement that faithfulness metrics have been…
Thibault Bañeras Roux, Jane Wottawa, Mickael Rouvier, Teva Merlin, Richard Dufour · Apr 30, 2026 · Citations: 0
Pairwise PreferenceAutomatic MetricsGeneral
However, they remain system-oriented, even when transcripts are intended for humans.
In this paper, we firstly present Human Assessed Transcription Side-by-side (HATS), an original French manually annotated data set in terms of human perception of transcription errors produced by various ASR systems.
The Generator is optimized to produce increasingly effective prompt injections, while the Defender is strengthened through preference-based training to maintain consistent reviews between clean and attacked submissions.
The resulting model is a fully decoder-only language model with 82.2M trainable parameters and no dense FFN sublayers, compared with a 103.0M-parameter dense GPT-style baseline used in the evaluation.
3.2903/26.85), while showing close zero-shot benchmark behavior under the evaluated setting.
Unlike standard NLP benchmarks, business idea evaluation relies on multi-dimensional criteria such as feasibility, novelty, differentiation, user need, and market size, and expert judgments often disagree.
This paper studies a methodological question raised by such disagreement: should an automatic judge approximate an aggregate consensus, or model evaluators individually?
Zhengxu Yu, Yu Fu, Zhiyuan He, Yuxuan Huang, Lee Ka Yiu, Meng Fang · Apr 24, 2026 · Citations: 0
Automatic MetricsGeneral
Individual agent capabilities have advanced rapidly through modular skills and tool integrations, yet multi-agent systems remain constrained by fixed team structures, tightly coupled coordination logic, and session-bound learning.
To fill this gap, we introduce OneManCompany (OMC), a framework that elevates multi-agent systems to the organisational level.
Weixu Zhang, Ye Yuan, Changjiang Han, Yuxing Tian, Zipeng Sun, Linfeng Du · Apr 24, 2026 · Citations: 0
Pairwise PreferenceAutomatic MetricsGeneral
In this work, we adopt a mechanistic interpretability perspective and hypothesize the existence of a sparse set of Preference Heads, attention heads that encode user specific stylistic and topical preferences and exert a causal influence on…
We introduce Differential Preference Steering (DPS), a training free framework that (1) identifies Preference Heads through causal masking analysis and (2) leverages them for controllable and interpretable personalization at inference time.
Yunho Kim, Quan Nguyen, Taewhan Kim, Youngjin Heo, Joonho Lee · Apr 24, 2026 · Citations: 0
DemonstrationsGeneral
Here we present Learning-Augmented Robotic Automation, a hybrid system that integrates learned task controllers and a neural 3D safety monitor into conventional industrial workflows.
We deployed the system on an electric-motor production line to automate deformable cable insertion and soldering under real manufacturing constraints, a step previously performed manually by human workers.
Emerging AI systems in behavioral health and psychiatry use multi-step or multi-agent LLM pipelines for tasks like assessing self-harm risk and screening for depression.
We present a statistical framework for multi-agent pipelines structured as directed acyclic graphs (DAGs) that provides an alternative to heuristic voting with principled, adaptive decision-making.
