SceneTeract is a verification interface that separates semantic action understanding from physical feasibility in indoor 3D scenes. It decomposes activities into atomic actions and performs explicit geometric checks to determine executability, providing diagnostic traces for failures. The system reveals widespread functional and accessibility issues in synthetic scenes, shows that existing VLMs over‑predict action feasibility, and improves VLM performance through post‑training with verifier feedback, with benefits that generalize to real‑world scenes.
By L\'eopold Maillard, Francis Engelmann, Tom Durand, Boxiao Pan, Yang You, Leonidas Guibas, Maks Ovsjanikov
The paper introduces BATON, a dual‑axis policy optimization framework for large language model agents that addresses two key aspects of reinforcement learning: intra‑trajectory feedback attribution and inter‑trajectory objective aggregation. BATON combines Bayesian Feedback Attribution, which builds a feedback‑conditioned posterior over sampled actions, with Trajectory Mass Normalization, which ensures equal optimization weight for complete trajectories. Experiments on ALFWorld, WebShop, and SearchQA demonstrate that each axis independently improves performance and that their combination yields the best results across different model scales.
By Yingxuan Zhuang, Binhe Yu, Jingxiao Yang, Ruopei Sun, Ziting Li, Cheng Tan, Xuhong Zhang, Jianwei Yin, Jintao Chen
arXiv:2609. 20519v1 Announce Type: new Abstract: As coding agents move from supervised code completion to unattended, around-the-clock exploration, their work expands from isolated predictions into long trajectories of reasoning, tool use, and feedback.
By Haozhe Liu, Tian Ye, Sensen Gao, Qihang Cao, Yitong Li, Mingchen Zhuge, Duomin Wang, Ruihua Zhang, Ping Luo, Jiawang Bian, Lei Zhu, Ligeng Zhu, Enze Xie, Song Han
RAFT is a retrieval‑augmented framework designed for troubleshooting agents that treats support cases as stateful, multi‑stage documents. It represents each closed case as a directed chain of timeline entries and retrieves at the entry level, matching intermediate states of an active case and returning the parent‑case trajectory anchored at the matched state. The authors evaluate RAFT on a synthetic benchmark and real Apache Jira issues, showing statistically significant improvements over vanilla RAG and GraphRAG baselines, and release the benchmark, implementation, and evaluation set.
By Mingxuan Zhang, Xiaowen Wang, Anupma Sharan, Zhengyi Chen, Chenyu Diana Zhang, Shanshan Yang, Chittibabu Pacharu
The paper introduces On‑Demand Attention (ODA), a decoding strategy that lets pretrained language models decide when to use global attention based on a lightweight recall head. ODA keeps the original model weights unchanged, only training the recall head, and can be implemented with GPU‑side conditional execution to reduce global reads. Experiments on Qwen, Gemma, and hybrid‑attention models show that ODA largely recovers performance lost by local attention while cutting the number of global attention operations.
By Haibo Feng, Ruiqi Liang, Hanyang Peng, Shiqi Yu
The study investigates how limited reading capacity and claim wording influence consensus outcomes in language‑model networks. By modeling message capacity as the number of messages an agent reads, the authors show that when agents read fewer than about 6.4 messages on average, a wrong consensus becomes unreachable. However, the wording of a claim—its inherent threshold—can override this effect, leading to incorrect consensus even when most agents start correct.
By Makoto Fukushima
The paper introduces a three‑layer checklist-and-judge framework to evaluate interpreter agents that mediate live conversation across languages. It assesses semantic, pragmatic, and cultural‑social dimensions—naturalness, intent, and social appropriateness—rather than just fidelity, in both single‑turn and multi‑turn settings. Extensive validation shows that conventional MT metrics miss failures in stronger interpreters, and that context, structured instructions, and cultural cues influence communicative success.
By Faiz Ghifari Haznitrama, Alice Oh
The paper reports the first large‑scale empirical comparison of AI‑agent and human online communities, analyzing 73,899 Moltbook and 189,838 Reddit posts across five matched communities. It finds that Moltbook shows extreme participation inequality (Gini = 0.84 vs. 0.47) and high cross‑community author overlap (33.8% vs. 0.5%). Linguistically, AI‑generated content is emotionally flattened, more assertive than exploratory, and socially detached, leading to community‑level homogenization that is largely a structural artifact of shared authorship. At the individual level, AI agents are more identifiable than human users due to outlier stylistic profiles amplified by their extreme posting volume.
By Agam Goyal, Olivia Pal, Hari Sundaram, Eshwar Chandrasekharan, Koustuv Saha
The study explores how neural agents develop dependency length minimization (DLM) in artificial languages using a recurrent neural network framework. By manipulating processing constraints such as listening noise, speaker capacity, and incremental sentence processing, the researchers find that DLM emerges only under incremental processing pressure, while other factors produce varied word‑order preferences. These results suggest that human cognitive processing limits may influence the emergence of DLM in language.
By Yuqing Zhang, Tessa Verhoef, Gertjan van Noord, Arianna Bisazza
The paper introduces Active Cross-View Object Geo-Localization (ActiveGeo), enabling mobile agents to actively select new viewpoints and decide when to stop to improve localization with fewer observations. It proposes the ActiveMoPT framework, which uses a three-stage training process: Multi-View Prompt-Preserving Adaptation, Trajectory-Guided Policy Initialization, and Cost-Aware Policy Refinement with GRPO. The authors also create a zero-shot test set, ActiveGeo-858, and demonstrate that ActiveMoPT outperforms prior methods on MoP-UAV and ActiveGeo-858.
By Shunyu Yao, Xiaohan Zhang, Zhuoran Yang, Haoqi Lai, Qi Ming, Xiaoxi Hu, Hui-Liang Shen, Si-Yuan Cao
The paper introduces the concept of xeno-interpretability, which studies internal distinctions in large language models that lack corresponding human concepts. It distinguishes between human‑interpretable and xeno‑semantic spaces, showing that LLMs possess a far larger internal representational space than can be captured by finite human descriptions. The authors propose an empirical program to identify and characterize these xeno‑representations, noting their potential to influence model behavior in ways that are not fully visible through human‑readable communication.
By F. Pierucci, M. Bracale Syrnikov, M. Prandi, M. Galisai, F. Giarrusso, P. Bisconti
The paper introduces Search-to-World, an end‑to‑end evaluation task that measures how well agentic systems can turn user requests into usable 3D worlds by combining request understanding, web visual‑content retrieval, and world delivery. It defines two metrics—Observed Retrieval Rate (ORR) and World Delivery Rate (WDR)—to separate the act of finding relevant content from successfully delivering a perceptually acceptable 3D world. The authors also present WorldSearcher, a reusable‑then‑reconstruction harness that connects existing search agents to world delivery, and a structured recovery controller that can revise queries or source videos when failures occur. Benchmarking with WorldSearcher shows that delivery success varies with the underlying agentic model and that observing relevant content alone does not guarantee world delivery; jointly training recovery agents improves both success and efficiency.
By Zixiao Gu, Yabo Chen, Xunzhi Xiang, Yu He, Haibin Huang, Chi Zhang, Yunbo Wang, Xuelong Li
The paper argues for a Foundation Model Operating System (FMOS) to virtualize foundation model interactions, similar to how operating systems abstract hardware. Current AI stacks are fragmented, with each framework embedding its own runtime for state, memory, budgets, and guardrails, leading to non-portable behavior and brittle governance. An FMOS would orchestrate knowledge across memory tiers, manage model selection and resource allocation, and enforce verification and policy, learning when to intervene or allow direct inference based on operational experience.
By Suparna Bhattacharya, Tarun Kumar, Cong Xu, Satish Kumar Mopur, Jiahao Li, Ashish Mishra, Aalap Tripathy, Annmary Justine Koomthanam, Martin Foltin, Ian Foster
The paper "What Do We Expect from LLMs? Mapping the Design of LLM Benchmarks" analyzes 14,767 arXiv submissions from 2022 to 2026 that introduce or update evaluation resources for large language models. It systematically maps changes in target systems, domains, evaluation materials, conditions, and scoring mechanisms, revealing a growing emphasis on action, interaction, and professional applications. The study also notes uneven development in model participation, with LLM-based scoring increasing in both agent and non-agent groups, while model-generated materials do not show a comparable rise.
By Chao Wang (Independent Researcher)
The study examines how large language model (LLM)–based graphical user interface (GUI) agents respond to digital nudges. Using Dual‑Process Theory, researchers tested 3,600 agents across six frontier models in an online shopping experiment and found that the agents were susceptible to both automatic (Type 1) and reflective (Type 2) nudges. The agents’ reasoning configuration moderated these effects in opposite directions: extensive reasoning reduced susceptibility to automatic default nudges but increased susceptibility to reflective social‑influence nudges, with the effect systematically varying by model scale.
By Haya Halimeh, Sascha Kaltenpoth, Kevin B\"osch, Oliver M\"uller
The paper presents a three‑phase pipeline that shifts semantic user profiling from per‑user to per‑transaction‑pattern inference, dramatically reducing LLM inference costs. The Resolve phase maps item names to abstracts, Profile generates categorical labels, free‑text attributes, and prevalence estimates per pattern, and Tag clusters these attributes into a searchable database. Evaluated on a public e‑commerce corpus, the pattern‑based approach matches the performance of a per‑user LLM while enabling deployment at a major Japanese bank with tens of millions of users and a three‑order‑of‑magnitude reduction in inference targets.
By Ryota Mitsuhashi, Tetsuro Morimura, Hirotake Ito
The paper introduces a digital‑twin framework that simulates opinion dynamics in real Twitter networks by assigning agents attributes such as persona, emotions, centrality, stubbornness, and influence, and using Mistral‑7B to update opinions based on memory and social exposure. Validation on COVID‑19 and U.S. election 2020 datasets shows the framework reproduces opinion trajectories, reducing prediction error by over 50% compared to classical baselines, and improves structural alignment and polarization dynamics. Ablation studies reveal that agent attributes, memory, and social exposure all contribute to predictive fidelity, with agent attributes being the most critical.
By Omran Berjawi, Giuseppe Fenza, Rida Khatoun, Sherali Zeadally
arXiv:2609. 20353v1 Announce Type: new Abstract: We study repeated contract design when a principal observes outcomes but not the actions that generate them.
By Rui Ai, David Simchi-Levi, Han Zhong
BioPhys-Bridge is a newly released benchmark dataset designed to evaluate language models on evidence‑grounded scientific reasoning within biophysical literature. Each of its 500 cases includes evidence blocks, stable IDs, quantitative values, units, equations, assumptions, mechanisms, and next‑step decisions, covering six biological domains and nine physical model families. The dataset enforces strict quality gates and has already been evaluated against several models, with DeepSeek‑V4‑Flash achieving the highest evidence‑ID F1 score of 0.360.
By Qingyang Xu
The paper evaluates three approaches—LLM-Z3 full replanning, IPyHOPPER hierarchical repair, and iTIMO local-revision—for revising travel itineraries after disruptions such as flight cancellations or hotel unavailability. Using two TREK-derived benchmark sets (500 single-disruption cases and 200 compound-disruption cases), the study compares effectiveness, plan stability, and computational cost. Results show LLM-Z3 with Gemini achieves the highest compound-disruption success, IPyHOPPER nearly matches single-disruption success while preserving more of the original itinerary, and iTIMO makes fewer edits but consumes more tokens, offering practical guidelines for balancing feasibility recovery, commitment preservation, and computational cost.
By Xiaofei Yuan, Yan Zhang, Shaobo Qiao, Huangleshuai He, Leyan Ni, Mingchen Ju, Lujia Yang, Sijia Xu, Yifu Tang, Zhengyi Yang