MAGMA-GEN is an on‑policy data‑generation pipeline that transforms ambiguous failures in hierarchical robotic manipulation into validated recovery supervision. It uses a privileged coach to hypothesize early decision‑level errors and proposes localized corrections, then retains only those candidates that improve downstream progress when re‑executed from the same state. This approach generates supervised examples from the agent’s own failure distribution, enabling improved task success and recovery without requiring per‑step human demonstrations.
By Loan Bernat (LAAS-GEPETTO), Matthieu Grard (LAAS-RAP), Ariane Herbulot (LAAS-RAP), Florent Lamiraux (LAAS-GEPETTO)
The paper investigates how agent harnesses—specifically planning guidance, execution organization, and completion verification—affect performance in retail and airline pilot tasks. By comparing fixed, task‑specific plans to shuffled policy text of equal length, the study finds that fixed plans improve success rates by about 7 percentage points, especially on complex tasks. A read‑only verifier rejects a majority of invalid episodes while incurring minimal cost, and its impact varies with the penalty for erroneous acceptance, often matching the full planning‑plus‑verification benefit at a lower cost.
By Yukun Zhang, Kemu Xu, Yishen Chen
Deep Noir is a framework that autonomously discovers optimal activation‑steering parameters in transformer models by leveraging Logit Lens convergence and causal head‑level attribution. It demonstrates significant performance gains across models ranging from 1B to 9B parameters, improving spam detection by up to 42 percentage points and SST‑2 sentiment classification by 13.1 percentage points without code changes. The approach also reveals that increased steering magnitude expands a predictable prompt‑injection attack surface, highlighting security implications for agent systems using steered classifiers.
By Frank E. Bobe III, Gregory D. Vetaw, Darshan W. Bryner, Matthew G. Cook, Jose L. Salas-Vernis
The study investigates how individual components of a coding harness—planning, action space, and context management—affect autonomous coding agents’ performance. By fixing the execution loop and varying these components across 176 settings on SWE‑Bench Verified and Terminal‑Bench 2.1, the authors find that context management is most valuable when context windows are tight, staging rule‑based elision before LLM summarization yields the best efficiency, planning serves as an accuracy scaffold for weaker models and a cost saver for stronger ones, and predefined tools help models with limited bash skills while bash‑capable models benefit from a bash‑only interface. Trajectory‑level analysis shows that context management lengthens execution paths, planning alters where trajectories terminate, and the action space determines code granularity, offering a modular framework for future harness design.
By Run-Ze Fan, Zihao Zhang, Simin Ma, Yebowen Hu, Shouju Wang, Kaiqiang Song, Fei Liu, Hamed Zamani, Xiaoyang Wang
The paper introduces the AR Fairness Metamodel, a structured framework for representing, analyzing, and comparing fairness scenarios. It incorporates key elements such as agents, resources, and their attributes, and supports both discrete and continuous fairness measures—including equality, equity, group fairness, individual fairness, the Gini index, the Theil index, Jain's fairness index, and a specific measure for Australia's Child Care Subsidy. The metamodel builds on the Tiles framework, offering modular components that can be connected to capture diverse fairness definitions, and includes formal proofs of relationships among group fairness, individual fairness, and envy‑freeness. An open‑source implementation of the Tiles framework is provided to facilitate practical fairness modeling and evaluation across various applications.
By Julian Alfredo Mendez, Timotheus Kampik
The paper introduces HALTER, a graph-based system that automates the reset and evaluation of long-horizon robot manipulation tasks. HALTER constructs a spatial scene graph from point clouds and vision models, uses an LLM to score rollouts, plan resets, and verify success, all without labeled success images. In experiments on a Franka arm, HALTER restores scenes in 76% of episodes, improves skill completion estimation, and reduces operator time by 72% compared to manual reset.
By Jing Jiang, Yue Yang, Xinkai Jiang, Gedas Bertasius, Daniel J. Szafir, Rudolf Lioutikov
The survey "From Models to Systems: A Comprehensive Survey of Efficient Multimodal Learning" reviews over 300 works on efficient multimodal learning (EML), proposing a structured taxonomy that spans model, algorithm, and system layers. It synthesizes how cross‑layer co‑design addresses the Efficiency‑Utility‑Privacy trade‑off and illustrates this through a case study of multimodal large language models. The paper also offers optimization blueprints for various domains, discusses a shift toward self‑regulating intelligence, and outlines open challenges for future EML research.
By Pan Wang, Siwei Song, Hui Ji, Siqi Cao, Heng Yu, Zhijian Liu, Huanrui Yang, Yingyan Celine Lin, Beidi Chen, Mohit Bansal, Xiaoming Liu, Pengfei Zhou, Ming-Hsuan Yang, Tianlong Chen, Jingtong Hu
The paper introduces DASE, a query engine designed to efficiently link unstructured data for multi-step reasoning tasks. DASE combines a multi-step reasoning model, a sparse materialized embedding-similarity join index (SemJI), and a co‑designed execution layer to perform multi‑attribute filtering, multi‑vector search, exact relational joins, and thresholded embedding‑similarity joins. In scientific discovery workloads, DASE outperforms traditional RDBMS, rerank, and vector‑database baselines by 6x to 46x in retrieval speed while maintaining comparable recall, and it serves as a high‑recall prefilter that reduces downstream LLM evaluation cost and improves accuracy on benchmarks such as SemBench E‑Commerce.
By Jiaming Liang, Haydn Jones, Jacob R. Gardner, Mark Yatskar, Zachary Ives
AURORA is a natural‑language‑driven framework that treats air‑ground scenario generation as a compilation process with verification. It introduces the Air‑Ground Scenario Graph (AGSG), a typed intermediate representation linking agents, missions, events, communication, and success conditions, enabling joint grounding, temporal planning, pre‑execution checks, runtime verification, failure localization, and bounded repair. The authors also present AURORA‑Bench to evaluate not only execution but faithful realization of requested interactions, showing that structured execution and runtime verification improve reliability and that explicit intermediate representations facilitate verifiable and repairable co‑simulation.
By Keshu Wu, Hao Zhang, Rui Gan, Xiangbo Gao, Xiaopeng Li, Zhengzhong Tu, Yang Zhou
DeltaSelect is an open‑source method that selects a small, cost‑effective set of tasks for A/B testing coding agents, using Pearson correlation to identify tasks whose single‑run results reliably reflect full‑benchmark performance. It maps fractional verifier outputs to a common score via linear regression and operates within a specified dollar budget, making it suitable for repeated baseline‑versus‑candidate comparisons during development rather than for model ranking. In a case study with gpt‑5.6‑luna, DeltaSelect reduced costs by 58.1% and improved calibrated scores across 13 evaluations.
By Nicholas J. Conn
The paper introduces FARSIGHT, a framework for evaluating the robustness and security of financial trading agents powered by large language models. It assesses agents on their resilience to market turbulence, such as flash crashes, and their vulnerability to three types of attacks: on information sources, on the agents themselves, and on agents acting as attackers. Applying FARSIGHT to 15 academic trading schemes reveals that 80% fail at least one robustness test and all exhibit security weaknesses, highlighting the risk of market-wide crashes from both accidental misjudgments and deliberate attacks.
By Mengxiao Wang, Nitesh Saxena
DataCanvas-EDU is an agentic framework that lets instructors guide the creation of synthetic datasets for business analytics courses. Instructors set teaching goals and desired patterns via conversation, and an AI agent writes generation code, verifies the data, and produces assignments, reference solutions, and rubrics. The process is organized into four phases—Plan, Create, Verify/Test Analysis, and Evaluate—to streamline case preparation and enable students to explore new patterns with AI.
By Bang An, Maria Hamdani, Joseph Fox
DeliveryGym is a 3D reinforcement learning environment that simulates continuous courier shifts, integrating multimodal tool interaction, persistent world dynamics, and trajectory‑based rewards derived from simulator events. It allows agents to learn how their decisions affect time, energy, and money across an entire shift, and it adapts future training shifts to the policy’s weaknesses while keeping evaluation fixed. Experiments on six models and 13 city maps show a significant gap between task execution and optimal sequencing, with RL improving Qwen3‑VL‑4B’s net income by 54.3% and adaptive training boosting test income by 16.5% over uniform sampling.
By Haoqiang Kang, Yiming Zhang, Yiyang Guo, Chuying Li, Jianzhi Shen, Tianruo Rose Xu, Xiaokang Ye, Lianhui Qin
AdaRepair-Mem introduces an adaptive experience retrieval framework for repository-level program repair, addressing three key limitations in existing memory-augmented methods: imbalanced episodic memory, non-monotonic success with increased memory, and phase misalignment of memory types. The framework employs coverage-aware retrieval, quality-aware selection, and stage-aware routing to better match repair contexts with relevant memories. Evaluations on SWE-Bench-Lite and SWE-Bench-Verified show improved performance on under-covered repositories, reduced noisy retrieval, and enhanced patch refinement support.
By Z. C. Luo, J. C. Guo, W. J. He, S. Y. Wang, J. C. Yu, F. M. Zhao, Y. Chen, T. Cao, L. Q. Liu, N. Zheng, W. Xu, J. Jiang, Z. M. Zhao
The paper introduces an automated pipeline that converts non‑critical driving scenes into safety‑critical scenarios by integrating computer vision, Large Language Models (LLMs), and Augmented Reality (AR). It detects and tracks road users, extracts safety features such as distance, velocity, motion direction, and Time‑to‑Collision (TTC), and evaluates scene criticality. Safe scenes are then modified by an LLM, which generates realistic collision‑inducing objects and behaviors that are overlaid onto the original scene using AR, achieving 97.52% safety classification accuracy on the nuScenes dataset and producing realistic scenarios like pedestrian crossings, rear overtaking vehicles, and sudden‑stop events.
By Noura Fady, Farah Khaled, Catherine M. Elias
STR-Agent is an LLM-driven framework designed for QoS-aware routing in Low Earth Orbit satellite networks. It integrates intent perception, tool-based execution, experience accumulation, and reflection-based policy adaptation to translate natural-language service requests into adaptive routing decisions. In simulations on a Walker-Delta constellation, STR-Agent reduces end-to-end delay by up to 60% compared with DQ-Dijkstra and improves intent-understanding accuracy from 45.4% to 92.45% after fine-tuning, with the Reflection Module providing additional delay reductions.
By Bowen Lu, Mugen Peng, Yaohua Sun, Hongyu Wang, Kerui Guo, Wenjia Xu
The paper introduces SPAR, a closed‑loop simulation platform that couples real‑time AUV control software with a higher‑level orchestration layer for fault injection, prompting, and evaluation of large language models (LLMs) in diagnosing and recovering from anomalies. SPAR enables ensemble testing of LLMs, comparing a frontier model with three locally deployable LLMs on a mass‑shift fault scenario across 480 trials, revealing that model choice significantly affects diagnostic accuracy. The study demonstrates that while the frontier model consistently ranks the correct fault mechanism among its top hypotheses, local models succeed mainly when they follow the full diagnostic procedure, and overall diagnosis and operational decisions appear decoupled in this dataset.
By Khalid Halba, Kylie Cooper, James G. Bellingham
The paper introduces prediction‑powered smoothing (PP‑S) and its taxonomy‑aware extension (PP‑TS) to improve point and interval estimates of domain‑specific AI performance when only a limited sample of labeled units is available. It also proposes a new design‑based cross‑validation score that is approximately unbiased for selecting between direct and smoothed estimators. Experiments on a curated benchmark and real‑world agent traffic show that the proposed methods outperform direct estimators in both accuracy and coverage, and that the new score matches the performance of an independent validation sample while providing more precise error estimates.
By Sho Kawano, Zehang Richard Li, Paul A. Parker
RetireOPD introduces a self-retiring on‑policy distillation method for agentic reinforcement learning. It first trains a skill‑conditioned teacher with environment rewards, then jointly trains a skill‑free student with RL and OPD, allowing the student to autonomously stop using the teacher when its performance aligns with the teacher’s. Experiments on Qwen2.5 models show significant gains in ALFWorld success rates and WebShop accuracy compared to RL baselines and the teacher itself.
By Yan Yu, Zhengxi Lu, Yizhou Liu, Yichen Pan, Aozhe Wang, Qipeng Chen, Hua Yang, Wenqi Zhang, Weiming Lu, Qianglong Chen, Yongliang Shen
TripScore is a benchmark and evaluation framework for large language models (LLMs) in travel planning, built from real user logs and calibrated with 1,468 pairwise judgments from 203 travel experts. It uses a hierarchical feasibility gate for format and commonsense checks, and a unified point-wise reward that combines soft quality and preference fulfillment. Experiments show that reinforcement learning fine‑tuning, such as GRPO, consistently outperforms other methods when evaluated with TripScore.
By Yincen Qu, Huan Xiao, Feng Li, Gregory Li, Hui Zhou, Xiangying Dai, Xiaoru Dai, Xuan Huang