MoMHa is a system that optimizes large language model harnesses across three objectives—accuracy, behavioural safety, and token cost—using a single‑phase joint‑reward proposer. It outperforms alternative strategies on seventeen domains, including synthetic suites and real‑world benchmarks, achieving higher joint scores and better safety while reducing token usage. The approach demonstrates that multi‑objective harness design can transfer effectively to unseen models and tasks.
By Subhojyoti Mukherjee, Md Mehrab Tanjim
arXiv:2609. 30971v1 Announce Type: new Abstract: Embodied agents offer a promising route to automating scientific experimentation, yet their progress is constrained by the lack of reliable and systematic evaluation environments.
By Maokai Qin, Chuan Qin, Qi Zhang, Dianyu Liu, Zirui Liu, Hongting Niu, Yuanchun Zhou, Hengshu Zhu
The paper introduces CodeHack, a library of code-based skills with natural-language descriptions designed to improve language agents in complex environments like NetHack. By allowing agents to invoke reusable skills instead of selecting individual actions, the study shows that skill-based agents nearly triple game progression and cut inference cost by 86% in zero‑shot settings, while still retaining the option to fall back on primitive actions. In reinforcement learning, skill-based agents learn faster, achieving a 7.2× larger average gain in dungeon level within the same training budget.
By Bart{\l}omiej Cupia{\l}, Jens Tuyls, Maciej Wo{\l}czyk, Davide Paglieri, Martin Klissarov, Benjamin Eysenbach, Piotr Mi{\l}o\'s, Karthik R. Narasimhan
OmouAI is an interactive deliberation system that combines large language models with computational argumentation to facilitate policy debates involving humans and simulated personas such as stakeholders, experts, or devil’s advocates. Each persona generates its own arguments, which are assembled into a shared argumentation framework that users can contest, add to, or revise, ensuring human oversight. The system evaluates arguments using deterministic argumentative semantics against external goals like the UN Sustainable Development Goals, providing faithful explanations and indicating how policy recommendations affect those goals.
By Stylianos Loukas Vasileiou, Antonio Rago, William Yeoh, Georgina Curto
The paper introduces a momentum-guided federated split distillation framework for personalized temporal edge intelligence. It presents TeRR-SAtt, a temporal reservoir student attention design that uses fixed reservoir representations, a lightweight temporal student, and personalized output modules. Additionally, it proposes AMGF, an anticipatory momentum-guided fusion mechanism that clusters clients via learning momentum and generates specialized teacher updates. Experiments on real-world smart‑building data show that TeRR-SAtt cuts edge training latency by 65.50%, inference latency by 44.70%, training memory usage by 18.40%, and inference CPU usage by 33.10% compared to baselines, while AMGF improves local learning by up to 35.31% in RMSE over global updates.
By Ahmed-Rafik Baahmed (LINEACT), Jean-Fran\c{c}ois Dollinger (LINEACT), Amine Brahmia (LINEACT), Mourad Zghal (LINEACT)
arXiv:2609. 31176v1 Announce Type: new Abstract: Repository-level issue localization aims to identify and rank the files and functions relevant to resolving a reported issue.
By Yunxiang Wei, Zhenyu Lei, Jundong Li
G2MAF is a test‑time refinement framework for offline multi‑agent reinforcement learning that applies a single globally normalized, projected critic gradient to adjust all agents’ actions while keeping them close to a frozen policy proposal. The method improves performance on 24 Multi‑Party Environment (MPE) and StarCraft Multi‑Agent Challenge (SMAC) benchmarks, achieving mean relative gains of 9.2% on MPE and 8.9% on SMAC, with only a 6% increase in inference latency.
By Guowei Zou, Haitao Wang, Guoxin Wang, Zhiquan Chen, Beiwen Zhang, Guojie Wang, Hejun Wu
The paper introduces LOHA, a context layout that compresses older tool observations into soft tokens while keeping the agent’s own turns and the last K observations in plain text, and ACD, a training method that distills full‑text predictions into this latent representation while anchoring behavior on plain text. This approach reduces context per call by up to 57% without significant loss in resolve rates, and improves instance throughput in single‑GPU serving. Experiments on SWE‑bench Verified show that K=3 yields a 43–57% compression with only modest performance impact, while larger windows favor task performance over compression.
By Zhensheng Zou (Peking University), Guoqing Wang (Peking University), Dan Hao (Peking University)
The paper introduces SeLATM, a framework that improves topic modeling by generating topics at the segment level and refining them through agentic feedback loops. This approach addresses limitations of LLM-based topic assignment methods, such as the inability to produce topic distributions, overly broad or narrow topics, and high resource consumption. Experiments on multiple datasets show that SeLATM reduces LLM resource usage while maintaining superior performance.
By Myeongjun Erik Jang, Antonios Georgiadis, Sae Young Moon, Fran Silavong
The paper introduces Steiner’s taxonomy of group tasks to study how multi‑agent large language model (LLM) teams scale on disjunctive versus compensatory tasks. By modeling agents as conditionally independent given the item, it shows that plurality voting converges to the modal answer while averaging converges to the item‑level bias. Experiments with 13 open‑weight models and up to 30 agents reveal that disjunctive tasks benefit from larger teams, whereas compensatory tasks like Fermi estimation see little improvement, highlighting that task structure and aggregation method fundamentally determine team scaling.
By Carolina Fortuna, Blaz Bertalanic
DeepEdu‑v1 is an AI‑tutoring system tailored for Vietnamese education that addresses data‑sovereignty and local curriculum alignment issues. It uses a long‑context inference engine to reduce retrieval calls and prefill latency by about 35%, and a self‑improving agentic layer that curates verified local knowledge without fine‑tuning. In deployment, DeepEdu achieves nearly twice the speed of standard vLLM serving and raises agentic accuracy from 70.0% to 79.5% on complex tasks, especially in financial reasoning and interactive‑agent benchmarks.
By Quang Nguyen, Hieu Nguyen, Hien Hoang, Toan Pham, Cong Tran, Nam Vu
The paper presents a pipeline for generating multi‑turn synthetic conversations and a self‑improvement loop that uses variance‑based contrastive optimization and a coding agent to refine planning and tool‑use in conversational recommendation agents. This approach improves agent quality by 8% over a manually optimized prompt and has been deployed at Spotify, where it accelerated development cycles. In production, the system achieved a 14% increase in user listening, a 5% rise in weekly active users, and a 5% reduction in skip rate compared to a prior session‑only experience.
By Enrico Palumbo, Alexandre Tamborrino, Victor Ode, Ben Lacker, Adri\`a Casas Escoda, Jeremy Hopple, Marcus Better, James Leoni, Hugo Galv\~ao, Hugues Bouchard, Mounia Lalmas, Jos\'e Luis Redondo Garc\'ia, Abenezer Abebe, Ann Clifton, Anton Blomberg, Henrik Lindstr\"om, Dani Doro, Christine Doig Cardet
The report examines how software architects view the growing use of AI development agents in their field. A focus group of 22 industry and academic participants discussed current practices, trust, validation, governance, and educational implications, concluding that decision‑making, accountability, and guardrail authoring remain human responsibilities. They introduced the concept of harness engineering—building systems that govern AI‑assisted creation—and identified criticality and cognitive debt as key factors for calibrating human oversight.
By Uwe van Heesch, Olaf Zimmermann, Christian Kohls
The article evaluates the Sola Security Brain, a purpose-built security intelligence layer, against a general-purpose coding agent (Claude Code) on 28 cloud‑security investigation tasks. The Sola Security Brain achieved 0.693 coverage versus 0.387 for the coding agent, a 79.2% relative gain, and outperformed the agent on 25 of 28 tasks while incurring far lower reasoning and cost per unit of coverage. The study also identifies a ‘sample‑and‑generalise’ pattern where the live agent reports universal negatives based on limited sampling, illustrating a potential efficiency trade‑off in cloud investigations.
By Leon Goldberg, Gal Engelberg, Eden Yavin, Elad Elouz, Ariel Zadok, Konstantin Koutsyi
The paper introduces Inquesto Score (IS), a protocol that measures voice‑agent reliability by calculating the percentage of calls that reach the caller’s goal without functional failure. IS defines explicit failure events and severity levels, evaluates timing, semantic, and state‑dependent failures using audio, scenario predicates, tool traces, and a pinned open‑model judge, and provides diagnostic views on behavior, acoustic robustness, identity handling, and speaker groups. The authors evaluate IS v0.1 on 30 scenarios, three acoustic conditions, four speaker groups, and 306 calls per agent across 13 configurations, demonstrating that reliable measurement requires evidence beyond transcripts and explicit treatment of deployment conditions.
By Massa Baali, Bhiksha Raj
The paper audits runtime failure monitors that use a model’s internal representations to predict failures in autonomous driving tasks. Across two tasks—online vectorized map generation with LaneSegNet and end‑to‑end planning with VAD—the authors find that frame‑level errors can be predicted with high AUROC scores using supervised latent probes. However, adding latent features to baseline monitors that use only observable inputs and outputs does not yield statistically significant improvements, suggesting that internal representations may not provide additional predictive value beyond what is already observable.
By Nikhil Kamalkumar Advani, Vishwajeet Shivaji Hogale, Saurav Kumar
The paper introduces MemProbe, a framework inspired by cognitive science to evaluate stability-plasticity tradeoffs in agent memory systems. It offers four experimental paradigms—interference, misinformation, consolidation strength, and reconsolidation window—to manipulate memory updates, preservation, and uncertainty. Using a 56-episode diagnostic suite, the authors evaluate six incremental memory systems, revealing that similar overall scores can mask distinct behavioral profiles in how memories are updated, preserved, attributed, and temporally organized.
By Jiaqi Ding, Guorong Wu
The paper introduces KNOWS, a benchmark for evaluating web agents that act as assistants by retrieving, synthesizing, and presenting information across complex, multi-step browser tasks. It outlines a task design rubric, evaluation protocol combining deterministic checks with LLM judgments, and reports that current agents achieve only modest success, with the best performing agent succeeding on less than 3% of tasks. The study highlights significant gaps in agents’ tool use, visual understanding, and long‑horizon reasoning.
By Alexander Gill, Md Farhan Ishmam, Xuyen Nguyen, Neha Bhat, Parker Henry DeYoung, Fateme Hashemi Chaleshtori, Nathan Stringham, Kenneth Marino, Ana Marasovi\'c
The paper introduces a three‑step framework for deploying multi‑UAV networks in threat‑prone environments. First, a threat‑aware K‑means algorithm determines the minimal number of UAVs and safe initial positions. Second, an optimal matching stage assigns UAVs to these positions to reduce energy use. Third, a threat‑aware multi‑agent twin delayed deep deterministic policy gradient (MATD3) algorithm dynamically optimizes UAV trajectories, power, and user associations, achieving zero safety violations and superior energy efficiency compared to other learning methods and heuristic baselines.
By Faisal Al-Kamali, Hussein A. Ammar, Francois Chan, James H. Bayes, Yasser Gadallah, Mohamed H. Ahmed
The paper introduces LIRSeg, a method that replaces explicit Chain-of-Thought reasoning in multimodal large language models with a compact set of learnable latent tokens for reasoning segmentation. LIRSeg is trained in two stages—spatial alignment and GRPO—while employing extreme-advantage sampling, decoupled exploration-stability updates, and latent diversity amplification to enhance token informativeness. Experiments show that LIRSeg improves segmentation accuracy and reasoning efficiency, achieving significant gIoU gains over the VisionReasoner baseline and reducing reasoning tokens by about 16×.
By Tianhang Guo, Yulin He, Wei Chen, Wenjuan Zhou, Yuhang Li, Xinbiao Gan