benchmarking

Existing software modeling datasets are often ad hoc constructions lacking rigorous quality assurance, leading to research findings that are difficult to reproduce, compare, and prone to bias. This work proposes the first benchmarking framework specifically designed for model-driven engineering, treating datasets themselves as first-class evaluation targets. By defining clear metrics for quality, representativeness, and task suitability, the framework establishes a unified platform that enables automated analysis of modeling datasets across multiple languages and formats. For the first time, this approach facilitates systematic evaluation of modeling datasets, substantially enhancing the reproducibility, fairness, and scientific rigor of research in the field.

Scientific software selection frequently suffers from non-reproducible benchmarks due to multi-library, multi-metric evaluation and dynamic evolution—such as the introduction of new algorithms or modifications to test cases and evaluation criteria. This paper addresses numerical integration over arbitrary 2D/3D domains with implicit or parameterized boundaries (cut-cell quadrature), proposing the first automated benchmarking framework that systematically integrates CI/CD engineering practices into scientific computing workflows. The framework unifies GitHub Actions, Docker, Python-based scheduling, Jupyter-based report generation, and semantically versioned result archiving. It supports automated configuration, execution, visualization, and historical result comparison. It achieves >90% automation for benchmark tasks and regression detection; reduces integration time for new libraries or algorithms by 70%; and enables precise attribution of performance deviations to specific code commits. The framework significantly enhances reliability, reproducibility, and evolutionary adaptability in scientific software evaluation.

Current evaluations of agent tool use often conflate workload specifications, action generation, and evidentiary criteria, lacking a unified and auditable framework. This work proposes an evaluation paradigm centered on “evidence admissibility gating,” which explicitly decouples workloads, drivers, and verification evidence through a shared evidence admissibility contract. The framework integrates diverse environments—including WebArena Verified, a subset of SWE-Gym, and MiniWoB++—and employs a standardized reporting pipeline comprising a universal workload adapter, declarative drivers, task manifests, event schemas, and replay/freeze strategies. It uniformly logs multidimensional metrics such as latency, invalid actions, and patching costs, enabling consistent differentiation of controller performance under identical workloads while ensuring relevance, reproducibility, and auditability in agent evaluations.

This work addresses the lack of systematic and rigorous performance benchmarking methodologies in programming language research, which has undermined the credibility of evaluation results. To remedy this, the paper introduces a closed-loop methodology—Measure-Explain-Test-Improve—that establishes, for the first time, a structured and reproducible workflow for performance assessment in the field. Integrating systematic experimental design, performance metric analysis, result interpretation, and iterative refinement, the approach emphasizes theoretical grounding and practical rigor at every stage. Its key contribution lies in enabling even researchers with limited empirical experience to conduct reliable and methodologically sound performance evaluations, thereby significantly enhancing the scientific validity and reproducibility of performance analysis in programming language research.

Existing evaluation of generative AI is hindered by the scarcity of high-quality benchmarks, whose manual construction is costly and time-consuming. Method: We propose the first automated benchmark construction framework powered by collaborative large language model (LLM) agents, decomposing benchmark creation into four sequential stages—planning, generation, verification, and evaluation—integrating task decomposition, agent coordination, human-in-the-loop feedback, and explicit constraint-satisfaction assessment. Contribution/Results: The framework significantly enhances data diversity and metric reliability. Leveraging it, we construct the first high-quality benchmark specifically targeting planning and constraint-satisfaction capabilities in text generation. We systematically evaluate seven state-of-the-art models, uncovering shared failure modes and fine-grained capability disparities. Our work establishes a scalable, reproducible paradigm for evaluating generative AI capabilities, advancing both benchmark methodology and empirical analysis.

Existing benchmarks for knowledge work evaluation largely adhere to traditional NLP task paradigms, failing to capture systems’ capabilities in real-world knowledge-intensive settings. This work proposes a three-step framework—explicitly defining work activities, establishing realistic test environments, and focusing evaluation on deliverable outputs—and derives 18 core knowledge work activities from the O*NET database. Innovatively integrating role responsibilities, local tool usage, and downstream usability into benchmark design, the approach establishes a coherent “work activity–test setup–scoring artifact” alignment. Validation through three case studies (GDPval, OfficeQA Pro, and APEX-SWE) exposes critical misalignments in current benchmarks between tasks, environments, and actual work objectives, offering a new paradigm for evaluating knowledge work systems in practical, application-oriented contexts.

This study addresses a critical gap in existing tool-calling evaluation benchmarks: the lack of validation of the evaluators themselves, which risks conflating assessment artifacts with agents’ true capabilities. Through a systematic audit of four prominent benchmarks—BFCL v4, τ2-Bench, LiveMCPBench, and MCP-Atlas—the authors conduct expert review of 496 tasks, replicate experiments, and perform trajectory-level analysis, revealing an 18.5% disagreement rate between automated evaluators and human judgment. Notably, LiveMCPBench exhibits a score variance of up to 18.9 percentage points upon re-evaluation, sufficient to overturn leaderboard rankings. To address these issues, the work introduces the first unified taxonomy of tool-calling evaluation failures, advocates for distinct measurement of tool invocation, task completion, and result verification, and releases Tool-Veritas—a configurable benchmark—and Harness Lab, an open-source evaluation platform.

Traditional benchmarks provide only aggregate scores, offering insufficient evidence to support reliable deployment decisions and thereby creating a disconnect between evaluation and action. To address this gap, this work proposes a “deployment-completeness” benchmarking framework, introducing novel metrics—evidence fibers, completeness curves, and certifiable proportions—alongside a systematic audit methodology comprising evidence fiber analysis, response ranking intervals, conformal coverage evaluation, and a certify-then-acquire decision pipeline. Empirical evaluation on benchmarks such as Tox21, Matbench, and JARVIS reveals that conventional approaches suffer a drastic drop in channel coverage to 10.07% under real-world deployment conditions. In contrast, the proposed method reduces error-driven deployment decisions to 0.027% on Tox21 and 0.128% on JARVIS, substantially enhancing deployment reliability.

Existing approaches struggle to effectively quantify the similarity and quality between synthetic and real data in evaluating tool-augmented agents. To address this gap, this work proposes SynAE, a novel framework that establishes the first multi-axis evaluation system tailored for multi-turn tool-use scenarios. SynAE introduces four fine-grained metric categories—assessing task instructions, tool invocations, final outputs, and downstream evaluation performance—to systematically measure synthetic data across dimensions of validity, fidelity, and diversity. Integrating natural language processing, trajectory modeling, and controllable generation techniques, the framework enables a reproducible evaluation pipeline and successfully identifies several representative failure modes in synthetic data generation. Empirical results demonstrate that such multidimensional assessment is essential for enhancing the reliability of agent evaluations.

This work addresses the reproducibility challenges posed by the rapid evolution of large models and high-performance computing systems, where existing benchmarks lack sustainable and automated evaluation mechanisms. To bridge this gap, the authors propose a user-agnostic continuous benchmarking framework that integrates principles from software engineering—particularly continuous integration—to establish an automated pipeline. This pipeline seamlessly combines systematic workflows with community-driven collaboration, delivering a reproducible and scalable benchmarking infrastructure for artificial intelligence and neuroscience research. The framework significantly enhances the sustainability, transparency, and collaborative efficiency of scientific evaluation in these fields.