Sign in

Author: DEROBS

  • SkillForgeBench: Evaluating Agent Reliability Grounded in Agent Skill Specifications

    SkillForgeBench: Evaluating Agent Reliability Grounded in Agent Skill Specifications

    SkillForgeBench: Evaluating Agent Reliability Grounded in Agent Skill Specifications

    arXiv:2610.14918v1 Announce Type: new Abstract: Evaluating an agent's ability to strictly satisfy diverse and procedurally complex requirements is essential to determining its practical reliability. Conducting such an evaluation demands agentic tasks that reconcile broad domain coverage with deep procedural complexity. Existing paradigms face a fundamental trade-off: expert-curated benchmarks achieve procedural depth but suffer from domain diversity bottlenecks, whereas crowdsourced benchmarks span diverse scenarios but lack execution rigor. To resolve this tension, we present SkillForgeBench, which leverages agent skills as unifying execution blueprints across three co-designed stages: (1) Resource Preparation uses domain-balanced skills to guide artifact retrieval, establishing broad workflow coverage; (2) Skill-driven Task Synthesis employs skills as procedural anchors to partition artifact roles, synthesize naturalistic task instructions, and package sandboxed scenarios for deep procedural execution; and (3) Holistic Evaluation Design translates skill-embedded requirements into fine-grained completion and process rubrics to enable strict requirement-level scoring. Comprising 842 sandboxed tasks across five major professional domains, SkillForgeBench reveals a steep capability hierarchy among eight mainstream agents, led by Opus 4.8 ($40.9\%$ strict completion). Requiring all constrains satisfaction uncovers severe capability bottlenecks across all model tiers that remain masked under individual rubric pass rates ($70.0\%$ to $85.7\%$ vs. $12.6\%$ to $40.9\%$). Crucially, under the external guidance from skills, top-tier agents obtain larger gains than low-tier agents (i.e., $\Delta 19.2\%$ for Opus 4.8 vs. $\Delta 5.8\%$ for Haiku 4.5). Finally, agents satisfy process rubrics far less frequently than completion rubrics (under $3\%$ for Haiku 4.5), proving that completion-only checks overestimate real-world agent reliability.

    Fallback brief based on the available RSS feed; no unsupported details have been added.

    Source: arXiv
    Original article: https://arxiv.org/abs/2610.14918

  • Predicting how model propensities generalize out of distribution

    Predicting how model propensities generalize out of distribution

    Predicting how model propensities generalize out of distribution

    arXiv:2610.14937v1 Announce Type: new Abstract: How do language models generalize from their training data? This is a central challenge for alignment: models reliably learn their training distribution given enough samples, but we want models that behave well when deployed in new contexts. We study out-of-distribution (OOD) generalization of propensities. Our central experiment uses a range of elicitation techniques — system prompting, few-shot learning, SFT on reference answers, and GRPO — to increase a target propensity in a model, then measures how it spills over onto others across the resulting cross-elicitation matrices. We find that spillover is highly correlated between models when the same elicitation technique is used, but our SFT and GRPO runs induce different spillover profiles. Elicitation of benign traits can sometimes generalize towards undesired traits, for example a model trained to claim that it is sentient often becomes more power-seeking. We further find that spillover partially resembles trait correlations found in humans and identify three clusters of traits that account for about half of the variance using exploratory factor analysis (EFA). Across many elicitation runs, we observe that models change in characteristic ways that don't depend on the targeted trait, a phenomenon we refer to as drift. Drift moves Qwen3.5-9B towards its base model. When investigating if LLMs are able to predict spillover, we find that Claude Opus 5.5 predicts spillover slightly better than the models themselves, but worse than data-driven approaches using alternative elicitation methods or models. Overall, no single dominant predictor emerges, and generalization remains hard to predict. Code and data are available at https://github.com/nielsrolf/propensity-spillover.

    Fallback brief based on the available RSS feed; no unsupported details have been added.

    Source: arXiv
    Original article: https://arxiv.org/abs/2610.14937

  • MARL-BattleGrounds: A Novel JAX-native Benchmark for Heterogeneous and Competitive Multi-agent RL

    MARL-BattleGrounds: A Novel JAX-native Benchmark for Heterogeneous and Competitive Multi-agent RL

    MARL-BattleGrounds: A Novel JAX-native Benchmark for Heterogeneous and Competitive Multi-agent RL

    arXiv:2610.14883v1 Announce Type: new Abstract: We present MARL-BattleGrounds (MARL-BGs), an easy-to-use JAX-native benchmark for heterogeneous and competitive multi-agent reinforcement learning (MARL), designed to be sample-efficient, computationally efficient, and researcher-centric. MARL benchmarks are typically simple and efficient, like SMAX, where winning reduces to unit micromanagement, or rich and costly, like Dota 2. MARL-BGs fills this missing middle. Its first mode, Team Deathmatch, is a simultaneous turn-based game between two teams of five heterogeneous agents. The five classes have distinct basic, passive, and ultimate abilities designed to synergize with allies and counter enemies. Our 52 hand-designed maps with diverse obstacles include a 12-map curriculum and ten held-out validation and test maps. MARL-BGs provides two canonical evaluation modes. First, eight endgame scenarios form a hard generalization challenge: each asks a team to win an unseen, nearly lost position against a strong reactive controller within 5-10 ticks. Second, a monthly round-robin tournament retrains each submission on equal compute, plays it against every member of the current "Big Nine" field, and publishes the Elo results on the official leaderboard. All entrants are publicly released as downloadable opponents. MARL-BGs records 11,192 metrics per game in 28 topics. We also include three strong scripted opponents, an analytical replay viewer, and off-the-shelf tools for self-play, cross-play, curriculum learning, and zero-shot coordination evaluation. We even provide an API for LLM-based agents and use it to evaluate Qwen3.5-9B against our strongest policy. In our BattleClient, humans can play alongside or against any mix of policies, enabling human-AI evaluation. We benchmark MARL-BGs with six additional deep MARL baselines, which we release. On one RTX 5090 with 1,024 parallel games, the simulator runs 50,000 5v5 game steps per second.

    Fallback brief based on the available RSS feed; no unsupported details have been added.

    Source: arXiv
    Original article: https://arxiv.org/abs/2610.14883

  • Probable but Not Valid: A Psychometric Audit of Decision Models as Synthetic Survey Respondents

    Probable but Not Valid: A Psychometric Audit of Decision Models as Synthetic Survey Respondents

    Probable but Not Valid: A Psychometric Audit of Decision Models as Synthetic Survey Respondents

    arXiv:2610.14912v1 Announce Type: new Abstract: Language models are increasingly used to create synthetic survey samples, also called silicon samples. Decision models are a new class of language models that return a probability for each point of a rating scale instead of text. We ask whether these models, or the newest generative LLMs, can stand in for human respondents on a psychometric survey. We test fourteen decision models from eleven providers and 47 generative LLMs on an open Lithuanian organisational-psychology survey (n=263, 68 items), with 100 real respondent profiles, the Psychometric Similarity Score (PSS) and statistical baselines. The best generative model, MiniMax M3, reaches PSS 0.813. This is level with a Gaussian copula fit only on other respondents (0.824) and below a copula fit in sample (0.927). A second sample of 100 real people scores below the best models, so at this size PSS cannot rank them against real respondents. In 5 of 10 product lines the newest generative release scores below an earlier release of the same line beyond resampling noise, and none gains beyond it. With the default rounded readout, 13 of 14 decision models score below the generative reference gpt-5.4-mini. Of the 47 generative models, 33 reproduce the pattern of item correlations better than any decision model. Averaged over respondents, the probability vectors are much closer to the human item marginals than rounded answers. They also rank respondents about as well as a profile regression fit on 163 real people. As forecasts of each person's own answers, however, every decision model is worse than the answer frequencies of other respondents, mainly because its average vector differs from them. The profile regression improves on these frequencies by only 1.0%, so the profile fields say little about how a person answers. Synthetic respondents built on such profiles should therefore aim at sample-level fidelity.

    Fallback brief based on the available RSS feed; no unsupported details have been added.

    Source: arXiv
    Original article: https://arxiv.org/abs/2610.14912

  • Bridging Risk Approximation Gaps in Model Predictive Task Sampling via In-Context Modeling

    Bridging Risk Approximation Gaps in Model Predictive Task Sampling via In-Context Modeling

    Bridging Risk Approximation Gaps in Model Predictive Task Sampling via In-Context Modeling

    arXiv:2610.14913v1 Announce Type: new Abstract: Reinforcement learning (RL) is essential for adaptive decision-making, e.g., robust Meta-RL and domain randomization (DR), and efficient post-training of large language models (LLMs), where tasks are commonly structured as Markov decision processes or prompts. A central bottleneck in all these settings is the cost of agent-environment interactions required for task selection and policy optimization. Model predictive task sampling (MPTS) has emerged as a promising approach to improve sample efficiency by querying informative tasks via a risk predictive model (RPM) trained on optimization history. However, existing RPMs suffer from two failure modes: (i) sparse optimization histories, which limit RPMs' generalization across the task space, and (ii) a one-step lag in risk approximation, which causes biased difficulty estimation as the policy evolves. This work formalizes these failure modes, derives lagged-risk decompositions that expose the resulting temporal bias, and recasts risk prediction as an in-context sequence modeling problem. The developed RPM integrates a one-step look-ahead mechanism with a temporal-weighted sliding window to mitigate data non-stationarity and scarcity, without requiring additional rollouts. Empirically, our approach yields more reliable difficulty estimates and consistent performance gains across robust Meta-RL, DR, and prompt curriculum RL post-training of LLMs.

    Fallback brief based on the available RSS feed; no unsupported details have been added.

    Source: arXiv
    Original article: https://arxiv.org/abs/2610.14913

  • PEAM: On-device Communication Agent for Personalized, Efficient and Autonomous Mobile Network Configuration

    PEAM: On-device Communication Agent for Personalized, Efficient and Autonomous Mobile Network Configuration

    PEAM: On-device Communication Agent for Personalized, Efficient and Autonomous Mobile Network Configuration

    arXiv:2610.14855v1 Announce Type: new Abstract: The growing deployment of compact large language models on smartphones is enabling autonomous on-device decision-making, such as emerging mobile GUI agents. However, many mobile decisions, such as network configuration when users switch apps or connectivity conditions change, must satisfy stringent latency constraints to preserve user experience. In this paper, we propose PEAM, an on-device LLM-based communication agent that autonomously generates personalized network-configuration rules executable within milliseconds. Specifically, PEAM first fine-tune the on-device LLM with a dataset constructed from manually annotated real-user data to integrate the communication knowledge into the local model. PEAM then operates in two phases. In the rule-generation phase, which is performed when the device is idle, such as overnight, PEAM analyzes historical network-configuration records in batches and reflects across batches to consolidate consistent evidence, reduce local bias, and derive robust configuration rules. To enable more accurate and stable rule generation over time, PEAM further constructs a user profile from high-confidence rules to capture the user's network preferences and incorporates this profile into each rule-generation cycle. In the rule-execution phase, PEAM is invoked when the communication context changes, such as when users switch apps or connectivity conditions change, and searches the rule library to identify and execute the appropriate rule. Experiments demonstrate that PEAM, based on a fine-tuned 4B model, achieves 99.13% correct network configuration decisions, while 92.50% of the rules generated by PEAM are valid, outperforming the traditional expert-rule based method by 4.41% and 84.50%, respectively.

    Fallback brief based on the available RSS feed; no unsupported details have been added.

    Source: arXiv
    Original article: https://arxiv.org/abs/2610.14855

  • WeatherBridge: Forecast-Aware Multi-Source Input Adaptation for Pretrained Weather Forecasting Models

    WeatherBridge: Forecast-Aware Multi-Source Input Adaptation for Pretrained Weather Forecasting Models

    WeatherBridge: Forecast-Aware Multi-Source Input Adaptation for Pretrained Weather Forecasting Models

    arXiv:2610.14863v1 Announce Type: new Abstract: Data-driven weather forecasting models are commonly trained on ERA5, yet operational deployment must use timely initial states from physics-based numerical weather prediction (NWP) systems whose distributions differ substantially from ERA5. Direct initialization with these states can degrade autoregressive forecasts, motivating input adaptation that keeps the pretrained forecasting model unchanged. Here, we introduce WeatherBridge, a unified, source-conditioned adapter that maps heterogeneous operational NWP initial states to ERA5 reference states without updating the pretrained forecasting model. WeatherBridge predicts the structured residual corrections with a Swin-UNet and injects source embeddings across multiple spatial scales, allowing the model to capture shared atmospheric structure and source-specific discrepancies. Its training objective combines latitude-weighted ERA5 reconstruction with One-step Rollout Alignment (ORA), which backpropagates one-step forecast error through a frozen FuXi model to optimize the adapted states for both initial state fidelity and downstream forecast compatibility. We evaluate WeatherBridge using daily GFS, HRES, and CMA initializations from 1 January to 22 November 2025 across six variables and lead times up to 240 hours. Relative to raw source initialization of FuXi, WeatherBridge reduces the mean RMSE of FuXi over the 6- 240-hour rollout for 17 of 18 source-variable combinations, with a median reduction of 10.2% and a maximum reduction of 46.3%. The adapter trained with frozen FuXi also improves FengWu forecasts on GFS inputs without retraining either the adapter or FengWu. These results demonstrate that forecast-aware input adaptation can improve forecasts from heterogeneous operational NWP inputs and transfer across pretrained ERA5-based forecasting models.

    Fallback brief based on the available RSS feed; no unsupported details have been added.

    Source: arXiv
    Original article: https://arxiv.org/abs/2610.14863

  • Beyond Outcome Feedback: Trajectory-Quality-Guided Experience-based Learning

    Beyond Outcome Feedback: Trajectory-Quality-Guided Experience-based Learning

    Beyond Outcome Feedback: Trajectory-Quality-Guided Experience-based Learning

    arXiv:2610.14877v1 Announce Type: new Abstract: Experience-based learning enables large language model (LLM) agents to improve over time by reusing knowledge from past executions. However, existing methods typically rely on final task outcomes as feedback for experience learning, overlooking the execution process and limiting their ability to evaluate and optimize experience based on trajectory quality. To address this limitation, we propose TQG-Exp, a trajectory-quality-guided experience-based learning framework that introduces process feedback. Specifically, we introduce Trajectory Quality Loss (TQL), which jointly considers task completion and process quality to provide a learning signal beyond final task outcomes. Based on this signal, we employ paired rollouts with and without retrieved experience to guide experience optimization. Representative trajectories are selected based on their outcomes and trajectory quality to extract task-level and task-type-level experiences. Meanwhile, differences in trajectory quality between the paired groups are used to estimate the contribution of retrieved experience and update its utility. The learned utility is then combined with semantic relevance to guide subsequent experience retrieval. Experiments on ALFWorld, WebShop, and ScienceWorld with two backbone LLMs show that TQG-Exp outperforms existing experience-based methods, improving Avg.@3 over the strongest baseline by 4.50 and 8.55 percentage points on DeepSeek-V4-Flash and Qwen3.5-Flash, respectively.

    Fallback brief based on the available RSS feed; no unsupported details have been added.

    Source: arXiv
    Original article: https://arxiv.org/abs/2610.14877

  • A Reference Architecture for Integrating Agentic AI into Enterprise Software: Responsibilities, Interactions, and Variability

    A Reference Architecture for Integrating Agentic AI into Enterprise Software: Responsibilities, Interactions, and Variability

    A Reference Architecture for Integrating Agentic AI into Enterprise Software: Responsibilities, Interactions, and Variability

    arXiv:2610.14818v1 Announce Type: new Abstract: Agentic AI moves generative AI applications toward systems that coordinate reasoning, enterprise context, model inference, and external actions. Integrating these capabilities into enterprise software introduces diverse architectural concerns, from orchestration, knowledge and memory, and model access to tools and actions, security, runtime infrastructure, and evaluation. Existing research and industry architectures address many of these concerns but differ in scope, abstraction, responsibility organization, interaction representation, and assumptions about which capabilities are required. This paper presents a technology-neutral reference architecture for enterprise agentic AI integration. We synthesize architectural capabilities from research publications, industry architecture guidance, and interoperability specifications by architectural purpose, organize them into ten responsibilities, and define boundary invariants, typed cross-responsibility interactions, and variation points that distinguish responsibilities from condition-dependent capabilities and replaceable realization mechanisms. We instantiate the architecture as an ESG portfolio-analysis and pre-trade compliance agent over an existing service-oriented ESG metric computation and reporting system, and complement the case with a mapping of an independently published agentic architecture. The case realizes seven of the ten responsibilities; different questions take different interaction paths, controls activate only under their conditions, and the probed boundary invariants hold. The independent architecture maps onto the proposed responsibilities without requiring an additional one. These results provide initial evidence for the architecture's organization and applicability rather than its completeness. The deployment and recorded results are available as a replication package.

    Fallback brief based on the available RSS feed; no unsupported details have been added.

    Source: arXiv
    Original article: https://arxiv.org/abs/2610.14818

  • EvoCowork: A Benchmark for Real-World Office Tasks

    EvoCowork: A Benchmark for Real-World Office Tasks

    EvoCowork: A Benchmark for Real-World Office Tasks

    arXiv:2610.14828v1 Announce Type: new Abstract: Office workflows are a demanding testbed for self-improving agents because their deliverables must be correct, natively editable, and mutually consistent. A central question is whether gains on feedback tasks reflect reusable procedures rather than task-specific fixes. We introduce EvoCowork, a 240-task benchmark for evaluating harness-level recursive self-improvement across four office modalities, with model weights held fixed. The benchmark covers presentations, documents, spreadsheets, and cross-office workflows, and scores submitted artifacts for content, native structure, formulas, cross-file consistency, and visual quality where needed. Its protocol separates feedback, revision selection, and final testing: methods learn from single-modality tasks, then are evaluated on unseen single-modality tasks and cross-office workflows under a shared initial harness and explicit budgets. In static evaluations of two models and four harnesses, the best configuration scores 42.25 overall and 30.62 on cross-office tasks. Starting from the same MSA harness with GLM 5.3, ACE and AHE improve in-domain scores by 4.90 and 2.64 points, respectively, but change cross-office scores by only +0.03 and -1.72. Thus, gains on the file types that provide feedback do not necessarily transfer to coordinated workflows. EvoCowork makes this distinction explicit while recording the capability, transfer, and cost of harness evolution. Code and task packages: https://github.com/DeepThinkingZhouLiu/cowork-evolution-benchmark.

    Fallback brief based on the available RSS feed; no unsupported details have been added.

    Source: arXiv
    Original article: https://arxiv.org/abs/2610.14828