model evaluation

The AI/ML community faces a severe reproducibility crisis, primarily driven by conceptual ambiguity in verification terminology—such as “reproducibility,” “replicability,” and “dependency/independence”—which undermines research credibility and scientific progress. To address this, we propose the first five-dimensional verification taxonomy, systematically defining core concepts—including reproducibility, dependency vs. independent re-executability, and direct vs. conceptual replicability—by clarifying their objectives, prerequisites, and evaluation criteria. Our framework integrates conceptual analysis, terminological standardization, and methodological modeling to yield a structured verification guideline. It enhances experimental rigor in study design, fosters consensus across the research community on verification practices, and significantly improves cross-team result reproducibility and outcome reliability.

This study addresses the uncontrolled quality of quality engineering (QE) artifacts—such as requirements specifications, test cases, and Behavior-Driven Development (BDD) scenarios—automatically generated by large language models (LLMs). We propose an iterative optimization framework integrating forward generation, backward generation, and rubric-guided scoring to enhance artifact quality along four dimensions: clarity, completeness, consistency, and testability. Our approach enables automated, quantitative, and reproducible quality assessment and improvement. Evaluated across 12 real-world projects, the method significantly improves output stability: it preserves high quality under high-quality inputs and substantially outperforms baselines under low-quality inputs. The core contribution is the first integration of backward generation with structured rubric-based guidance, establishing a closed-loop, artifact-centric quality enhancement paradigm for QE.

Large language models (LLMs) deployed for industrial test generation face critical reliability challenges due to rapid model iteration, leading to outdated evaluations and compromised production trustworthiness. Method: This paper introduces the first continuous evaluation framework for LLM-based test generation tailored to industrial settings. It pioneers a “continuous evaluation” paradigm integrating technical metrics (e.g., code coverage) with engineering metrics (e.g., maintainability, expert ratings), while systematically addressing real-world issues including data leakage and irreproducible results. The framework integrates industrial toolchains (e.g., SonarQube), supports dynamic test-case selection, robust prompt engineering, and auditable measurement infrastructure. Contribution/Results: A longitudinal empirical study at LKS Next demonstrates that the framework accurately tracks LLM capability evolution, identifies key bottlenecks impeding industrial deployment, and effectively enables trustworthy integration into DevSecOps pipelines.

Current evaluation practices for supervised learning models are often misleading due to an overreliance on single aggregate metrics, which neglect the alignment among data characteristics, task objectives, and real-world application contexts. This work reframes model evaluation as a context-dependent, decision-oriented process and systematically investigates—through controlled experiments—the impact of dataset properties, validation strategies, class imbalance, and asymmetric error costs on evaluation outcomes. Leveraging diverse benchmark datasets, multiple validation protocols, and multidimensional performance measures, the study uncovers common pitfalls such as the accuracy paradox, data leakage, and metric misuse. It proposes a structured evaluation framework explicitly aligned with operational goals, offering principled guidance for developing more robust, reliable, and trustworthy supervised learning systems.

This work addresses the efficiency bottleneck of manual reproducibility reviews in safety-critical domains such as the Internet of Things and cyber-physical systems, which hampers research transparency and deployability. The paper presents the first systematic framework leveraging large language models (LLMs) to automate reproducibility assessment by integrating natural language understanding, code generation, sandboxed environment auto-configuration, and rule-guided flaw detection. This approach enables reproducibility scoring, automatic execution environment setup, and identification of methodological flaws. Experimental results demonstrate that the proposed method achieves over 72% accuracy in reproducibility judgment, automatically constructs executable environments for 28% of runnable artifacts, and attains F1 scores exceeding 92% across seven common categories of methodological defects, substantially enhancing both the efficiency and quality of reproducibility review.

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 study addresses the underexplored engineering challenges in existing machine learning evaluation frameworks, where operational issues and their root causes have lacked systematic investigation. To bridge this gap, the work formally establishes evaluation engineering as a distinct research direction within software engineering. Through an empirical analysis of 57 frameworks and a comprehensive categorization of 16,560 reported issues across a newly proposed five-stage workflow model, the study reveals that 41.4% of problems originate in the specification phase, while 61.7% of classified issues stem from missing functionality, inadequate documentation, and insufficient input validation. The findings yield a structured taxonomy of evaluation-related problems and provide empirical evidence to inform the design and improvement of robust evaluation systems.

This study addresses a critical limitation of existing DORA metrics, which rely solely on first-order statistics and thus fail to capture the distributional characteristics of software release cadence or distinguish teams with markedly different release regularity. To overcome this, the work introduces second-order statistics into the DORA framework for the first time, proposing a novel Delivery Consistency (DC) metric based on the coefficient of variation of inter-release intervals. It further constructs an eight-prototype Delivery Health Matrix to enable multidimensional diagnosis and targeted intervention for software delivery rhythms across platforms. Validation using real-world data spanning 120 weeks from four platforms—including Jira, GitHub, and Firebase—demonstrates that the approach effectively identifies teams sharing identical DORA ratings yet exhibiting divergent release patterns, uncovering underlying organizational or process constraints common to such teams.

This work addresses the limitations of existing reproducibility assessment methods, which rely on manual annotations and thus lack scalability and authentic supervision signals reflecting real-world reproduction challenges. The authors propose the first scalable evaluation framework that leverages GitHub user-submitted issues as natural supervision, enabling large-scale assessment of large language model (LLM) agents’ ability to identify paper-to-code reproducibility issues without human annotation. By integrating language understanding with code context analysis, the approach enables non-execution-based detection of reproducibility barriers. Experimental results demonstrate that the best-performing LLM agent identifies at least one semantically relevant reproducibility issue—aligned with those reported by humans—in approximately 90% of the evaluated papers, exhibiting strong performance in both failure detection and semantic localization.

Existing demonstration data filtering metrics often achieve high performance on defect detection benchmarks such as AUROC but do not necessarily improve downstream behavior cloning policy performance. This work introduces controllable structural defects into the LIBERO pick-and-place benchmark to systematically evaluate the impact of various filtering methods on policy training. Experiments reveal a significant disconnect between AUROC and actual policy performance, with most metrics confounded by factors such as episode length. The study advocates evaluating filtering approaches based on policy success rate rather than detection accuracy, demonstrating that an optimal filtering strategy achieves a 90.0% success rate—approaching the 93.3% attained by an oracle using clean data—whereas methods yielding high AUROC scores result in the worst policy performance (13.3%). The authors open-source the testbed and evaluation pipeline.