Benchmark Validity

Follow this topic Ask about this topic

Momentum

15 papers in the last four weeks, up 114% on the four weeks before. 0.1% of all new papers.

Jul 13Week of Sep 28

Latest papers 117

Jul 30, 2026cs.AI

Safety, or Just Capability? A Validity Audit of Agent-Safety Benchmarks

Agent-safety benchmarks measure different behaviors, and their scores get quoted interchangeably as an agent's safety. We treat four of them (R-Judge, InjecAgent, AgentHarm, AgentDojo) as measurements to be validated, running each under its official implementation and author-provided scorer on up to 22 models, with MMLU and GPQA measured by us under one protocol as a capability composite. The metric is the first problem. On any binary trace-judgment benchmark scored by

F_1

, an ``always positive'' policy attains

F_1 = 2π/(1+π)

; on R-Judge that is

0.690

, above five of the 21 models that actually discriminate. The three broad-coverage benchmarks then rank the same 18 models differently, and the trade-off behind that disagreement is a small-panel artifact: R-Judge specificity against AgentHarm safety correlates

-0.64

at

n{=}7

and

+0.02

at

n{=}18

, and a quarter of random size-7 subsets reach

|ρ| \geq 0.5

around that near-zero value. Held-out validity turns on which outcome you pick. Capability predicts task success (

ρ{=}{+}0.60

) but correlates negatively with misalignment safety (

ρ{=}{-}0.44

,

n{=}21

). On their paired

n{=}20

panel, the corresponding contrast is

Δ{=}{-}1.00

(95% CI

[-1.48, -0.49]

,

p<0.001

), and it survives leave-one-organization-out and organization-clustered bootstrap analyses. On an expanded 41-model panel, the misalignment correlation weakens to

-0.16

(95% CI

[-0.54, +0.22]

) and jailbreak strengthens to

+0.34

, though neither change is significant. \mbox{AgentHarm} shows the strongest held-out association,

ρ{=}{+}0.72

with three-template jailbreak safety after controlling capability. But both instruments score harmful compliance, so this is evidence of convergent validity rather than general safety. Naming the benchmark, metric, target behavior, and model panel is the minimum a safety claim needs.

Youting Wang, Xiao Han, Dingyan Shang +2

LLM Safety Evaluation Benchmark Validity AI Agent Safety Benchmarks

Jul 29, 2026cs.AI

Automated Transcript Analysis for Detecting Flaws in Agentic Benchmarks

Capabilities of frontier models are often assessed using agentic benchmarks. To trust these results, benchmarks must accurately measure what they claim to and be free from invalidating flaws. Previous manual audits of benchmarks such as SWE-Bench-Verified have uncovered several validity issues in transcripts. However, manual review is difficult to scale, and it is unclear whether automated methods can reliably surface flaws that compromise benchmark validity. In this paper, we developed AI scanners to detect four types of validity issues: ground truth access, tool failure, guessing vulnerability, and answer format ambiguity. We produced grading rubrics for each to instruct human labeling, and evaluated the scanners against human labels on a held-out test set of Inspect Evals benchmarks. Our scanners identified several verified quality issues in five widely used benchmarks, including cases unlikely to be caught by random manual inspection. Not all cases were identified, and scanner performance varied substantially across benchmarks, criteria and models. We highlight several open challenges to be addressed to improve scanners for stronger quality assurance claims, including broader standardization gaps in the evaluation field that degrade scanner performance. Together, these results serve as a proof of concept for using automated transcript analysis to audit benchmark quality more broadly.

Jeff Mohl, Nelson Gardner-Challis, Magda Dubois +6

Independent · UK AI Security Institute · MARS +4

Benchmark Auditing AI Agent Benchmarks Benchmark Validity Automated Evaluation

Jul 28, 2026cs.AI

When benchmark inferences do not compose: Projectibility in AI evaluation

An AI benchmark result rarely reaches a consequential claim in one step. Evaluators generalize it to further cases, interpret it as evidence of capability, extrapolate it to new tasks, transport it to another system or site, and combine it with assumptions about human review and downstream consequences. Validity-centred approaches require evidence for each claim. This paper makes explicit and operationalizes a problem those approaches leave to the analyst: warranted links don't automatically make a warranted chain. The target of one study may not be the source of the next; system, population, outcome, or conditions may change at the interface; and shared data or model lineage may make apparently independent support dependent. Projectibility concerns whether a bounded extension from observed to unobserved cases is warranted. Goodman supplies the problem of rival extensions; argument-based validity supplies an architecture for testing them. The contribution is an interface audit for distributed AI evidence: typed source and target descriptions, and a procedure separating endpoints that never meet from endpoints that meet while warrant fails to cross. A legal-research case shows how benchmark evidence and a deployment study can each be sound while remaining parallel. A known-truth demonstration shows why aggregate stability can erase distinctions a later projection requires. The resulting projectibility audit diagnoses unsupported joins in benchmark-to-use arguments.

Brett Reynolds

Humber Polytechnic & University of Toronto

Benchmark Design Benchmark Validity

Jul 27, 2026cs.AI

Success Is Not Self-Explanatory: Auditing Success Provenance in Agent Evaluation

A correct answer can conceal why an agent succeeded. Once agents change their information state during evaluation, correctness no longer distinguishes intended reasoning from answer acquisition. Outcome evidence and exposure detection do not establish whether success depended on an acquired target; we call this missing evaluation object success provenance. AcquaBench audits it through matched CLEAN, GOLD, and SHAM value substitution on four standardized surfaces with joint qid-clustered analysis. CLEAN retains benchmark-authorized information. GOLD makes the correct target available. SHAM preserves source structure and exposure opportunity but substitutes a matched incorrect value. GOLD minus CLEAN measures the total score response to correct-target availability; GOLD minus SHAM tests whether that response tracks target correctness beyond matched source exposure. In D0, GOLD exceeds SHAM by 19.1 to 25.9 percentage points, showing that success follows the correct value. In D2, GOLD still exceeds SHAM under distributed sufficiency while coloc no longer transfers as a high-score marker, with AUROC 0.376 and 0.142. Behavioral dependence can thus persist beyond this probe's intended observation unit. In model comparison, a supported 5.0-point CLEAN score gap compresses to a raw GOLD difference of -0.6 points without establishing rank inversion. Agent benchmarks should report success together with whether the evaluated information state supported it.

Jingkun Luo, Da-Tian Peng

Northwestern Polytechnical University

Data Provenance Benchmark Auditing AI Agent Auditing Agent Evaluation AI Agent Benchmarks Benchmark Validity

Jul 26, 2026cs.LG

Charging Phase Health Indicators for Battery State-of-Health Estimation: A Systematic Comparison of CC, CV, and Combined Approaches under Cross-Battery Validation

Accurate State-of-Health estimation is essential for safe battery operation and cost-effective maintenance. Although numerous health indicators have been derived from constant-current (CC) and constant-voltage (CV) charging phases, their effectiveness under realistic cross-battery validation remains insufficiently studied. This work addresses this gap through a systematic comparison of CC-only, CV-only, and combined indicator sets using rigorous Leave-One-Battery-Out (LOBO) validation on the NASA battery aging dataset. Four CV-phase indicators and CC phase duration are evaluated individually and in combination. Results show that the combined CC+CV approach achieves the best performance (R2 = 0.874), confirming that CC and CV phases capture complementary degradation information. Moreover, a 119% performance gap is observed between standard 5-fold cross-validation and LOBO validation, indicating that conventional evaluation overestimates practical accuracy. Based on these findings, practical guidelines are provided for indicator selection under data and computational constraints.

Huy Hoang Le, Kim-Anh Nguyen

a R&D Department, PowerMore Ltd., Viet Nam · b Faculty of Electrical Engineering, University of Science and Technology, The University of Danang, Viet Nam

Benchmark Validity Lithium-Ion Batteries

Jul 23, 2026cs.AI

What AI Red-Team Evaluations Can and Cannot Prove

Red-team evaluations of AI models support some claims and not others, and the boundary between the two is calculable rather than merely a matter of judgment. We define the evidential ceiling of an evaluation as the largest factor by which one result can move belief under a fixed testing budget, derive it in closed form for the benchmark null result, and use it to locate that boundary exactly. We find that above a calculable harm rate, a benchmark of modest size certifies a category to a stated evidentiary standard, and a clean sheet is then the stronger of the two possible observations, outweighing a single reproduced failure. Below that rate, no passive benchmark of feasible size provides the specified evidence of safety under the fixed scoring rule and approximately independent trial structure. The crossing between the two regimes has a closed form. The bound is not specific to benchmarks: written in terms of a procedure's hypothesis conditioned elicitation rates, it covers adaptive and automated red teaming as well, and shows that discrimination between the hypotheses rather than attack success is what determines evidential worth. Auditing eight evaluation suites against the boundary, we find that current benchmarks are adequate for high-frequency harm categories and several orders of magnitude short for rare, catastrophic ones. Safety benchmarks are not uninformative. They are informative about a specific and computable set of propositions, and the discipline they need is to state which.

Bandana Kaur

APIsec Research Labs

LLM Safety Benchmarks LLM Red Teaming AI Safety Evaluation Benchmark Validity

Jul 13, 2026cs.AI

Good Benchmarks

Good tasks are correct, solvable, verifiable, well-specified, and hard for interesting reasons. The best tasks describe a real problem an experienced practitioner would recognize, in language a practitioner would use, with tests that verify the outcome rather than the approach.

Ivan Bercovich

Benchmark Design Software Engineering Benchmarks Benchmark Validity

Jul 9, 2026cs.CL

When the Judge Changes, So Does the Measurement: Auditing LLM-as-Judge Reliability

An LLM-as-judge score can move even when the candidate responses stay fixed, simply because the evaluator has changed. We treat this evaluator-replacement ambiguity as a measurement-validity problem. Across four judgment datasets, we compare two upgrade paths available in practice: scaling Qwen3 dense judges from 1.7B to 32B parameters and moving across MiniMax M2-M2.7 released APIs. The main pattern is that judge upgrades are not interchangeable: only Qwen3 1.7B to 4B gives a robust adjacent gain, while MiniMax adjacent releases do not. Stronger judges reduce but do not remove position and verbosity bias. Repeated-sample juries add little when errors are correlated. Structured debate can move decisions substantially, but without parser and fallback logs those shifts cannot be attributed to deliberation. We argue that LLM-as-judge reports should include dataset slices, bias probes, error-dependence estimates, and protocol audit trails.

Zongyou Yang, Yinghan Hou, Xiaokun Yang

Dyson School of Design Engineering, Imperial College London London, United Kingdom · Department of Electrical and Electronic Engineering, Imperial College London London, United Kingdom

LLM-as-a-Judge LLM Auditing Language Model Bias Evaluation Benchmark Validity

Jul 8, 2026cs.CL

From Text to Parameters: Predicting Item Parameters from Embedding Regularization with Reliability and Design Ceilings

Newly developed items must ordinarily be field tested before their psychometric properties are known, creating a cold start problem for item calibration. Predicting item parameters from features is a long standing measurement problem dating back to the Linear Logistic Test Model; modern text embeddings now automate the design matrices traditionally specified by hand. We propose an evaluation framework combining regularized regression on item text embeddings, repeated cross validated R squared reported with its resampling standard deviation, and two performance upper bounds: a reliability ceiling derived from parameter standard errors, and a design ceiling derived from simulation based power calibration. Applying this framework to a mathematics item bank (EEDI) and a medical licensure benchmark (BEA 2024), we find that item difficulty is highly predictable from text (repeated cross validated R squared = 0.53, or about 57% of its reliability ceiling), whereas discrimination and pseudo guessing appear less predictable. However, evaluating these results against our ceilings reveals that this apparent hierarchy stems from target reliability rather than text signal strength: text uniformly recovers 57 to 63% of the reliable variance across difficulty targets, whereas the 3PL pseudo guessing parameter has a reliability ceiling near zero, making it an unviable target at current precision. On BEA, embedding based regression matches leaderboard RMSE despite explaining almost no variance, highlighting the critical need for scale free metrics and explicit ceilings in benchmarking. Finally, we show that a single train and test split can inflate apparent accuracy by 0.1 to 0.15 in R squared, underscoring the necessity of repeated cross validation for calibration support applications and future benchmark construction.

Shi-Ting Chen, Jinsong Chen

Faculty of Education, The University of Hong Kong

Text Embedding Evaluation Benchmark Validity Item Response Theory

Jul 3, 2026cs.CV

Towards Standardized Light Field Quality Assessment: Hybrid Subjective Benchmarking and Objective Metric Evaluation

Benchmarking immersive media coding solutions, especially in the standardization context, requires reliable and reproducible subjective quality assessment (QA) procedures, along with objective quality metrics that remain accurate across different distortion types. This paper presents a standardized workflow for light field QA, developed and deployed in the context of JPEG Pleno standardization activities, which integrates benchmark generation, a hybrid subjective evaluation, and objective metric analysis into a common workflow. The benchmark is designed to encompass not only traditional coding-only artifacts but also distortions that arise in processing pipelines in which light field encoding is accompanied with view synthesis and reconstruction techniques. A hybrid subjective method is proposed enabling fine-grained assessment by combining reference-anchored quality rating with targeted pairwise refinement in perceptually ambiguous regions. The reliability of subjective scores is verified using statistical consistency analyses between observers of two cohorts. Finally, a large set of objective metrics is systematically evaluated in terms of global prediction accuracy, local agreement in ambiguous quality regions, and robustness across distortion families. The results show that several metrics achieve strong agreement for coding-only stimuli, but their performance consistently drops when view synthesis distortions are included. The analysis further highlights the importance of view-pooling strategy in the design of future light field quality metrics. The work provides a reproducible and standardization-ready framework for fine-grained light field QA, while identifying key limitations of current objective metrics under emerging coding pipelines. The subjectively annotated dataset is publicly available at https://plenodb.jpeg.org/lfqa/objectivecfp.

Saeed Mahmoudpour, Mylene C. Q. Farias, Gi-Mun Um +6

Dept. of Electronics and Informatics, Vrije Universiteit Brussel, Belgium · IMEC, Kapeldreef 75, Leuven, B-3001, Belgium

Pairwise Comparison Benchmark Validity Image Quality Assessment

Jul 2, 2026cs.AI

Meta-Benchmarks for Financial-Services LLM Evaluation

Public LLM leaderboards optimise for global average performance and do not capture the specific cognitive demands of financial-services work: a model that leads on MMLU-Pro may underperform on document-grounded compliance reasoning, and a coding leader may handle multi-turn customer interactions poorly. We present a meta-benchmarking framework that organises 452 publicly reported benchmarks into 41 O*NET Generalized Work Activities and aggregates those into 38 BIAN banking business domains spanning sales, operations, risk, and support work. A multiplicative weighting scheme (discrimination x coverage x recency), computed over a rolling model window, rewards benchmarks that still separate the best models, are widely reported, and remain in active use, suppressing saturated legacy tests automatically. These weights scale the K-factor in a pairwise Elo tournament, producing cross-benchmark-comparable work-activity scores without raw score normalisation; business-domain scores are weighted averages of the constituent work-activity Elos. We demonstrate the framework on a point-in-time public snapshot covering 288 models across 25 organisations as of June 2026, and describe the methodology, full taxonomy, design decisions, and limitations with the aim of making the approach reproducible for institutions facing similar selection and governance challenges.

Blair Hudson

Commonwealth Bank of Australia

Financial Services LLM Evaluation Benchmark Design Benchmark Validity Finance

Jul 1, 2026cs.LG

Auditing the Audit: Five Failure Modes in Benchmark-Validity Audits

Governance frameworks ask AI providers and auditors for documented evaluation evidence, and perturbation-based construct-validity audits are a common form of that evidence. We argue the audits are themselves fragile: their conclusions can be silently manufactured by implementation details that readers cannot see in the reported numbers. We name five classes of pipeline failure and demonstrate each in a self-audit over safety benchmarks and open-weight instruction-tuned models. Under a unified six-point due-diligence gate, every cell lands in a non-confirmatory bucket, and no cell reaches confirmatory. The evidence here is a single two-model, five-benchmark case study, and F1--F5 is an illustrative, deliberately non-exhaustive starting taxonomy -- not a comprehensive partition of audit failures. We position the gate as a withholding and disclosure protocol for assurance-grade evidence, supplementary to (not a replacement for) classical construct-validity evidence, and not as a route to benchmark-validity verdicts.

Yanhang Li, Zhichao Fan, Zexin Zhuang

Northeastern University, USA · University of Illinois Urbana-Champaign, USA · Southern Methodist University, USA

AI Assurance Benchmark Auditing AI Safety Evaluation Benchmark Validity

Jun 30, 2026cs.LG

Validating Causal Abstraction Metrics on Simulated Complex Systems

A central goal of science is to produce valid explanations of complex systems: high-level causal accounts that faithfully reflect the behavior of lower-level mechanisms. Yet no consensus exists on how to measure whether a proposed high-level explanation is actually valid. We introduce a benchmark of ten complex systems spanning both discrete and continuous state spaces, as well as static and dynamical regimes, each equipped with consensual ground-truth causal explanations and invalid contrastive conditions. Within a unified causal abstraction framework, we systematically evaluate over thirty candidate metrics drawn from observational, functional, information-theoretic, and causal families. Our results show that only the latter reliably discriminates valid from invalid abstractions, and only when incorporating faithfulness testing over unmapped variables. Building on these findings, we introduce the Causal Abstraction Error (CAE), a continuous validity metric with an explicit faithfulness test, which passes all discrimination tests across every system and can converge with as few as 30 sampled interventions. We offer it as a general-purpose metric for the discovery and validation of high-level explanations.

Maxime Méloux, Tiago Pimentel, François Portet +1

Université Grenoble Alpes, CNRS, Grenoble INP, LIG · ETH Zürich

Causal Abstraction Benchmark Validity Causal Explanation

Jun 29, 2026cs.LG

Exploring Differences Between Tabular Enterprise Data and Public Benchmarks

Tabular data dominate the landscape of data science, increasingly attracting innovative machine learning models and tailored benchmarks. Yet, little is known for enterprise data, where tables constitute the backbone of business operations. To broaden the benchmarking landscape for business applications, this work aims to actualize the characteristics of enterprise data by providing an analysis of data statistics and performance measurements of tabular models such as TabPFN, TabICL and ConTextTab. Through our analysis, we find enterprise data markedly differ from tabular benchmarks and we demonstrate that a tabular model that performs well on typical tabular benchmarks may perform poorly on real world enterprise data -- and vice versa. This lack of generalization underlines the need for additional benchmarks with enterprise-grade characteristics.

Myung Jun Kim, Maximilian Schambach, Frank Essenberger +2

SAP SE, Germany.

Benchmark Design Tabular ML Benchmark Validity Tabular Prediction

Jun 29, 2026cs.AI

EvalSafetyGap: A Hybrid Survey and Conceptual Framework for LLM Evaluation-Safety Failures

LLM evaluation and AI safety face a shared measurement problem: benchmark scores, reward-model signals, and reported safety metrics can improve while the latent properties they are meant to represent remain difficult to verify. This paper combines a hybrid survey - a systematic search paired with narrative synthesis and separately tracked grey evidence - with a conceptual framework and a structured ten-model audit. The synthesis spans eight evidence streams: benchmark validity, dynamic evaluation, LLM-as-judge reliability, safety evaluation, jailbreak/refusal robustness, reward hacking, mechanistic interpretability, and governance/auditability, covering 2018-2026 evaluation-safety measurement work. We introduce EvalSafetyGap as an organizing hypothesis for comparing evaluation-side and alignment-side proxy failures under optimization pressure, using Goodhart's Law together with two constructs we develop here - an Instability Decomposition and an Alignment Trilemma - as tools for generating testable comparisons. The audit shows how conclusions shift when capability, behavioral safety, and governance are measured separately. In this sample (

n = 10

), the association between capability and sustained adversarial robustness is statistically indeterminate using the displayed Table 3 inputs (Pearson

r = +0.232

,

p = 0.520

), and the apparent open-closed safety gap is modest, driven mainly by governance and disclosure rather than behavioral robustness, and sensitive to how a single borderline model is classified; attempt-budget results are protocol dependent. Because the public evidence uses heterogeneous protocols, the audit is diagnostic rather than rank-generating. The contribution is a shared vocabulary and evidence map to support dynamic evaluation, transparent source reporting, multi-attempt safety measurement, and auditable alignment practice.

Buğra Alperen Uluırmak, Rifat Kurban

Erciyes University, Türkiye. · Abdullah Gül University, Türkiye.

AI Alignment LLM Safety Alignment LLM Safety Evaluation Benchmark Validity

Jun 28, 2026cs.AI

Faults in Our Formal Benchmarking: Dataset Defects and Evaluation Failures in Lean Theorem Proving

Benchmarks for LLM-assisted theorem proving in Lean are often treated as intrinsically reliable because every solved instance comes with a machine-checked proof. However, the kernel only checks that a proof establishes a \emph{formal} statement; it does not verify that the statement faithfully encodes the intended informal problem, nor that evaluation harnesses are robust to trivial or adversarial solutions. We audit five widely used Lean theorem-proving benchmarks and their forks, using corpus-scale static checkers to surface 4,833 findings, including 398 mechanically certified issues such as counterexamples, vacuous theorems, and unsound axioms. We also document semantic defects such as missing hypotheses, problem simplification, incomplete or incorrect translations, and Lean-specific specification hazards. Beyond dataset construction, we survey evaluation-time failure modes and show, on corrected subsets, that defects can both inflate and deflate reported prover scores. We propose a fault taxonomy, a suite of automated checkers and recall-oriented semantic audit prompts, and release standards to guide the creation of formal math datasets and to make evaluation more reproducible and trustworthy. Our checkers, audit prompts, and corrected dataset snapshots are available at https://github.com/Shashi456/atp-checkers.

Pawan Sasanka Ammanamanchi, Siddharth Bhat, Stella Biderman

Eleuther AI · Cambridge

Benchmark Design Benchmark Validity Lean Theorem Proving Automated Evaluation

Jun 28, 2026cs.AI

Pooled Leaderboards Hide System-Specific Winners: A Reporting-Protocol Audit of Offline Root-Cause Analysis Benchmarks

Offline root-cause-analysis (RCA) benchmarks commonly rank methods by a single pooled top-1 accuracy across multiple subsystems, and engineers often read the pooled winner as a recommendation for their own subsystem. We audit that reading on three public RCA benchmark families -- OpenRCA, RCAEval, and PetShop -- covering 11 subsystems and 778 matched scoring units. To keep pairwise comparisons on identical cases, the main analysis retains four methods or comparators with complete coverage: BARO, a CD-1min adapter, max-

|Z|

, and per-service alert-count. All six pairwise comparisons show subsystem-level effects of both signs, every random-effects 95% prediction interval crosses zero, and case-level interaction tests reject exchangeability in 5 of 6 pairs. Leave-one-system-out selection picks the lower-scoring method on up to 5 of 11 held-out subsystems, with regret reaching 24.8 pp on RCAEval / Sock-Shop. We release a 320-line audit module; given a matched RCA benchmark score table, it recomputes the same per-subsystem stability checks alongside pooled scores.

Lining Hu, Ting Liu, Yuzhuo Fu

School of Electronic, Information and Electrical Engineering Shanghai Jiao Tong University Shanghai, China

Pairwise Comparison Algorithmic Auditing Benchmark Auditing Software Engineering Benchmarks Benchmark Validity Root Cause Analysis

Jun 26, 2026cs.SE

Building to the Test: Coding Agents Deliver What You Check, Not What You Requested

Benchmarks are widely used to evaluate task completion by Large Language Models (LLMs), but this approach has accumulated construction-validity problems, and a passing score may not show whether the requested task was delivered. We study both problems. In a controlled code-as-spec setup, two production Copilot CLI agents (claude-opus-4.7, gpt-5.5) re-implement a React Fluent-UI data table in Angular as a reusable library under a hidden 222-test Playwright oracle across 18 runs and three oracle-availability conditions. Alongside the score, we run a mechanical library audit and check each verdict with a no-op ablation. Without the oracle, the library is present but unfinished, revealed by scores. With the oracle in the loop, the score reaches near-perfect, but from a demo holding the tested behavior directly, the library left dead or absent. We call this building to the test; the broader disposition behind both we call validation self-awareness. The agent does not, on its own, validate what it ships as a user would. Prevalence remains an open question across other agents, signals, and model families. Beyond benchmark scores, dispositions like validation self-awareness merit research attention.

Yanuo Ma, Ben Kereopa-Yorke, Ben Schultz

Microsoft

Reward Hacking AI Coding Agents LLM Agent Evaluation LLM Agent Verification Benchmark Validity Code Generation Evaluation

Jun 23, 2026cs.AI

Life After Benchmark Saturation: A Case Study of CORE-Bench

When a benchmark's accuracy saturates, it is often retired and replaced with a more challenging version. We show that this approach privileges accuracy and misses the opportunity to study six other key dimensions of agent performance: construct validity issues such as shortcuts, out-of-distribution generalizability, efficiency, reliability, the relative importance of the model versus the scaffold, and uplift from human-agent collaboration. We use CORE-Bench Hard, a benchmark for computational reproducibility of scientific code, as a case study to demonstrate that measuring agents along these dimensions yields meaningful insights into agent performance even after accuracy saturates. First, we surface threats to construct validity in CORE-Bench Hard that are difficult to anticipate with less capable agents. We introduce an improved benchmark, CORE-Bench v1.1, and an out-of-distribution task suite, CORE-Bench OOD. Second, we find that despite accuracy saturation, CORE-Bench v1.1 remains useful for measuring efficiency, reliability, model performance, and scaffold performance. Finally, we conduct a small-scale randomized experiment to measure uplift from human-agent collaboration on real-world computational reproducibility tasks. We find a statistically significant speedup by about a factor of two -- likely underestimated due to one-fifth of human-only reproductions reaching the time limit before completing -- and describe various other findings. Together, our contributions present a more rigorous alternative to the dominant accuracy-centric evaluation paradigm.

Nitya Nadgir, Sayash Kapoor, Kangheng Liu +11

1Independent · 2Princeton University · 3UC Berkeley +1

OOD Generalization AI Agent Evaluation Benchmark Validity Scientific Reproducibility

Jun 22, 2026cs.LG

One Ruler: A Same-Hands Re-Evaluation of Bivariate Causal Direction on Tuebingen, with a Parameter-Free Compression Baseline

Headline accuracies on the Tuebingen cause-effect pairs are routinely compared across papers even though each is measured under its authors' own protocol -- different pair subsets, weightings, model-selection, and decision rates. We argue this is the wrong comparison and run the right one: a same-hands re-evaluation in which every method is run by us on the identical 102 pairs, with one strict rule -- no tuning and a decision forced on every pair. As a clean reference point we introduce a deliberately minimal baseline: sorted-conditional compression, which feeds quantized, sorted, first-differenced data to an off-the-shelf compressor (bz2) and has zero fitted parameters. Under the common ruler the ranking differs sharply from the literature. Our baseline reaches 74.7% weighted accuracy (p = 3.7e-7); on the same 100 pairs that SLOPE is evaluated on it scores 76.0%, a 1.2-point gap below the authors' own forced-decision SLOPE (77.2%) that is well inside noise (McNemar p = 0.39). A faithful re-run of RECI lands at 70.7% -- inside the original authors' reported error bar, not the 77.5% often quoted (which we trace to a mis-copied cell). SLOPE's published 82.4% is a decided-subset figure: scoring the authors' own stored output only on the pairs its significance test chose to answer reproduces 81.7%. Under the common ruler the methods cluster in the low-to-mid 70s and the zero-parameter compressor ties the strongest of them. We document the mechanisms that inflate published figures (test-set model selection, significance-gated abstention) and contribute two further results: compression score magnitude is a model-free confounding flag (p = 2.8e-68), and a pre-registered falsification test fails in an instructive way that bounds the method's theoretical interpretation. Code, pre-registrations, and per-pair outputs are released.

Wietse Stienstra

State of the Art — Compression Lab

Causal Discovery Benchmark Validity Causal Inference

Jun 19, 2026cond-mat.mtrl-sci

Computational references are not experiments: pre-registered validation of machine-learned sodium-cathode voltages

Machine-learning screens for battery materials are trained and judged almost entirely against computed reference voltages, and those references carry their own systematic errors. We report a case in which this matters quantitatively: our own screening stack (a graph-network voltage screen, a prior-art triage layer, and a local PBE+U bench) fails pre-registered validation against experiment-anchored literature values. Verdict thresholds, failure modes, and the primary metric were committed before analysis. On an operator-audited set of known Na-ion cathodes (n = 6 after one documented exclusion; verdict unchanged at n = 7), the raw held-out mean absolute error was 0.67 V, the pre-registered conservative metric, the upper 95% confidence bound of the cross-validated bias-corrected error, was 1.09 V, and the residual was strongly voltage-dependent (r = -0.94), so no additive calibration is valid. On the two compounds where prediction, database reference, and experiment could all be compared, the Materials Project PBE+U reference sat about 0.54 V below measurement: the reference, not the model, dominated the error. A prior-art screen found at least 70% of the targeted Na substitution space already published. We retire the screen, bound what "verified" means for our DFT ledger, and pre-register a calibration audit of it against four benchmark Li couples.

Krishna Teja Vepa

Independent Researcher†

Materials Property Prediction Virtual Screening Model Calibration Benchmark Validity Computational Materials Science

Jun 19, 2026cs.CL

The Metanym Game: A Self-Contained, Self-Consistent LLM Peer-Community Benchmark for Structural Intelligence

The metanym game is a competitive word game for LLMs that measures structural intelligence against established cognitive-science constructs. No content is given in advance; the contestants create all of it -- a new kind of analogy test, analogical production falsifiable sentence by sentence, with no fixed test set to leak into training (contamination-resistant by construction). In the council-of-peers benchmark, the contestants also rate each other's creations. We introduce the first spectral solution, to our knowledge, to the wicked problem of benchmarking LLMs' factual accuracy without golden keys or oracle models: one singular value decomposition of the evaluators' ratings matrix yields their competence as both generators and judges of true statements at once. Competence on the subjective criteria comes from each judge's rating consistency as the yardstick shifts. The factual rating correlates with GPQA Diamond at Pearson r = 0.92. Scored separately, making and judging dissociate -- judging is the scarcer skill: the strongest generators are middling judges, the sharpest judge a mid-pack generator. To scale, the strongest players form a council that does the official benchmarking; its seats are contestable -- a stronger model earns one on the benchmark's own rating. The benchmark is entirely self-contained and self-consistent, a stable gauge over time.

David Nordfors

LLM Evaluation LLM-as-a-Judge Synthetic Benchmark Generation Benchmark Validity

Jun 18, 2026cs.AI

Beyond Static Leaderboards: Predictive Validity for the Evaluation of LLM Agents

Agent benchmarks are growing fast, but no single benchmark touches more than four or five of the dimensions that deployment exposes. This paper aggregates the largest coordinated deep-dive of one MCP-based industrial-agent benchmark to date: fourteen parallel implementation studies covering new asset classes (including a multi-modal visual extension), alternative orchestrations, retrieval strategies, reasoning modes, infrastructure optimizations, and evaluation-methodology probes. Consolidating those studies with seven prior agent benchmarks, we argue that aggregate-score leaderboards systematically underspecify deployed-agent evaluation. Rankings derived from aggregate scores do not transfer to out-of-distribution settings; recent public-to-hidden competition retrospectives provide direct empirical evidence of this rank instability. We propose ranking configurations by predictive validity, the correlation between in-sample and out-of-sample rank, rather than in-sample mean, and report a twelve-tier measurement apparatus that exposes the deployment-relevant dimensions HELM and its agent-era successors collapse. The position is operationalized through three falsifiable out-of-distribution criteria with explicit thresholds; existing evidence partly supports it but is too thin to confirm. We close with a pre-registered pilot design and a field-level vision for what the next generation of agentic benchmarks should report.

Dhaval C. Patel, Kaoutar El Maghraoui, Shuxin Lin +58

OOD Generalization LLM Agent Evaluation AI Agent Benchmarks Benchmark Validity

Jun 17, 2026cs.LG

MassSpecGym in the Wild: Uncovering and Correcting Evaluation Pitfalls in AI-Driven Molecule Discovery

Reliable benchmarking is critical for developing machine learning models for tandem mass spectrometry (MS/MS) based molecule discovery. Subtle issues in experimental design and model evaluation procedures can degrade the trustworthiness of such benchmarks and lead to erroneous conclusions. We conduct a thorough review of model evaluation issues in the recent MS/MS machine learning literature, using the standard MassSpecGym benchmark suite as a case study to illustrate the impact of these issues. We find evaluation issues in at least 17 of 26 papers reporting MassSpecGym benchmark results in the first year of its adoption. We isolate three classes of failures: (i) data leakage, (ii) shortcut learning, and (iii) implementation bugs and metric divergence. Through extensive experimentation and code replication, we quantify the impact of these issues and show how they corrupt the evaluation standards MassSpecGym was designed to enforce. We distill our findings into recommendations generalizable to MS/MS challenges, benchmarks, and custom evaluation setups. We also release MassSpecGym v1.5, an implementation of our recommendations in the MassSpecGym benchmarking suite which addresses the failure modes identified in this audit. MassSpecGym v1.5 is publicly available at https://github.com/pluskal-lab/MassSpecGym.

Hongxuan Liu, Roman Bushuiev, Ivy Lightheart +12

1Massachusetts Institute of Technology · Institute of Organic Chemistry and Biochemistry of the Czech Academy of Sciences · 3Czech Institute of Informatics, Robotics and Cybernetics, Czech Technical University in Prague +2

Benchmark Design Benchmark Contamination Benchmark Validity Mass Spectrometry

Jun 17, 2026cs.CL

Reliability without Validity: A Systematic, Large-Scale Evaluation of LLM-as-a-Judge Models Across Agreement, Consistency, and Bias

LLM-as-a-Judge has become the dominant evaluation paradigm for language models, but judge validation in practice relies on exact-match agreement, a metric that does not correct for chance and systematically overstates discriminative ability. We present the largest systematic evaluation of LLM-as-a-Judge to date: 21 judges from nine providers across MT-Bench, JudgeBench, and RewardBench, evaluated under three protocols (agreement, consistency, bias audit) over 118 runs and approximately 541,000 individual judgments. Four findings emerge, consistent across the full cohort, including the April 2026 frontier: kappa deflation between exact match and Cohen's kappa is universal (33--41 pp on MT-Bench), judge rankings shift by up to 14 positions across benchmarks, high test--retest reliability (>0.95) coexists with severe position bias (>0.10) in two production-deployed judges (instantiating a consistency--bias paradox), and verbosity bias is small (<0.011) across our cohort under a single pairwise rubric. We distill these into a Minimum Viable Validation Protocol.

Justin D. Norman, Michael U. Rivera, D. Alex Hughes

UC Berkeley School of Information 102 South Hall Drive, Berkeley CA, 94708

Inter-Rater Reliability LLM Evaluation LLM-as-a-Judge LLM Reliability Language Model Bias Evaluation Benchmark Validity+1

Jun 14, 2026cs.CL

LLM Judges Have Dark Current: A Psychometric Datasheet for LLM-as-a-Judge Evaluation

LLM-as-a-judge systems are now routinely used for open-ended model evaluation, where human preference annotation is costly, slow, and difficult to reproduce. Yet these judges are often reported as scalar accuracy, win-rate, or agreement devices. We argue that a judge should instead be reported as a measurement instrument. We introduce a Judge Datasheet protocol that measures dark current under true-vacuum inputs, stable cross-sensitivity to same-quality surface variation, positional false preference, target sensitivity on a controlled quality ladder, and the criterion or operating point induced by tie instructions. The direction-stability decomposition reveals that apparent Delta0 preference can be stable surface response or disguised position bias. In a three-judge open-weight case study, Llama-3.1-8B shows high dark current and presentation-conflicted Delta0 behavior, Qwen2.5-14B is vacuum-clean and target-sensitive but mixes stable and positional over-discrimination, and Qwen2.5-32B is vacuum-clean with low stable cross-sensitivity and low positional false preference. A strict tie criterion eliminates Qwen32B Delta0 false preference but absorbs marginal Delta1 target signals into ties while preserving Delta5 sensitivity. The results show that prompting moves the criterion, not the resolution. We do not claim that the downstream mechanism hypothesis that motivated this work is confirmed; the contribution is a metrological protocol for measuring the measuring device before downstream claims are made.

Hiroyasu Usami, Keisuke Hara, Ayato Tsuboi +1

Department of Computer Science, Graduate School of Engineering, Chubu University, Kasugai, Aichi 487-8501, Japan · Mitsubishi Heavy Industries, Ltd., Research & Innovation Center, Heat Transfer Research Department, Takasago, Hyogo 676-8686, Japan

LLM Evaluation LLM-as-a-Judge Language Model Bias Evaluation Benchmark Validity Bias in LLM-as-a-Judge

Jun 10, 2026cs.LG

Crossing the Validation Crisis: Cross-Validation Reduces Benchmarking Variance Surprisingly Well

Modern machine learning progresses through empirical work, benchmarking new methods to evaluate relative performance. However, the statistical variability inherent to evaluation - exacerbated by the stochastic nature of many algorithms - often makes performance estimation unreliable due to the limited test samples available, leading to a validation crisis in which genuine advances are difficult to discern. In this work, we show that cross-validation improves markedly confidence when evaluating and comparing learning algorithm performances. We introduce the concept of sample gain, which quantifies the virtual data augmentation achieved by using multiple cross-validation splits to reduce benchmarking variance. Experiments on both synthetic and real-world datasets (histopathologic scans and NLP fine-tuning) demonstrate that multiple splits can substantially improve the reliability and stability of performance estimates, with diminishing returns often setting in later than expected. We also introduce a procedure to dynamically early-stop cross-validation by estimating from the first few folds if subsequent folds will bring large sample gains. Our findings highlight the value of pushing cross-validation on available samples to achieve robust and reliable benchmarking.

Célestin Eve, Gaël Varoquaux, Thomas Moreau

MIND Team, Université Paris-Saclay, Inria, CEA, Palaiseau, France · SODA Team, Inria, Palaiseau, France · Probabl

Benchmark Validity Variance Reduction

Jun 8, 2026cs.LG

ERBench: A Benchmark and Testsuite for Equation Discovery Algorithms

Equation discovery aims to automate the discovery of scientific models in the form of mathematical equations from data. Technically, equation discovery is implemented by symbolic regression algorithms. Performance of symbolic regression for equation discovery is measured along two dimensions: Prediction accuracy on test data, and recovery of known groundtruth formulas. For standard regression, accuracy is typically measured on in-domain test data, for instance, by splitting a data set randomly into training and test data. While this makes sense for in-domain interpolation, which is the common goal in ordinary regression, it can be a misleading proxy for true model discovery and generalization. The obvious alternative is to measure out-of-domain accuracy. However, obtaining challenging out-of-domain test data is a non-trivial problem. Therefore, we focus on equation recovery for evaluating symbolic regression algorithms for equation discovery. The rationale is that symbolic regression algorithms that perform well in recovering known groundtruth formulas are good candidates to perform well in unknown equation discovery. Existing benchmarks for symbolic regression include equation recovery tasks, however, with only a small number of groundtruth formulas that are publicly known. Moreover, these benchmarks place less emphasis on evaluating the robustness of algorithms in terms of their behavior under changing dimensionality, sampling size, sampling distribution and sampling domain. This, however, is of central importance to practitioners wanting to discover equations for modeling natural phenomena, since data is almost certainly noisy and comes from diverse domains, distributions, and sample sizes. To fill this gap, we introduce the Equation Recovery Benchmark (ERBench), a new evaluation framework designed to rigorously assess algorithms explicitly targeting the task of equation discovery.

Paul Kahlmeyer, Henrik Voigt, Michael Habeck +1

Department of Computer Science University of Jena Inselplatz 5, 07743 Jena, Germany

Benchmark Design Scientific Discovery Benchmark Validity Symbolic Regression

Jun 3, 2026physics.chem-ph

SC3: The Multi-Solvent Solubility Challenge and Benchmark

Solubility prediction is a standard benchmark in computational chemistry, yet multi-solvent models which reportedly approach the experimental-noise ceiling (i.e. the aleatoric limit) are not yet reliable enough to be deployed. We argue that this gap is partly artefactual: published benchmarks differ in curation policies, evaluate on count-weighted RMSE that hides failure on tail-heavy solvent distributions, and treat the widely cited 0.6-0.8 log S inter-laboratory figure as the aleatoric ceiling even though it reflects worst-case, not expected, disagreement. We introduce SC3, a multi-solvent solubility benchmark built on BigSolDB v2.1 with three contributions: (i) a reproducible curation pipeline yielding 101,535 measurements over 1,327 solutes and 206 solvents, with a recalibrated aleatoric floor of 0.106 log S-roughly 6 times tighter than the conventional figure; (ii) nested Gold/Silver/Bronze consensus tiers with per-point standard deviation, three leakage-checked splits, and a multi-solvent metric suite (PS-RMSE, Z-RMSE); and (iii) a 31-model benchmark across six families, whose best Bronze PS-RMSE sits at 5 times the aleatoric limit, and we observe this is a gap unclosed by any deep alternative tested. We perform three follow-on analyses: data scaling, transfer from quantum-chemistry solvation energies, and feature-level attribution, which demonstrates that calibrated per-point uncertainty is a reusable infrastructure for diagnosis beyond point prediction.

Vansh Ramani, Har Ashish Arora, Dhairya Kuchhal +4

Indian Institute of Technology Delhi, India · Kurnakov Institute of General and Inorganic Chemistry RAS, Russia

Benchmark Construction Benchmark Validity Molecular Property Prediction Uncertainty Calibration

Jun 2, 2026cs.RO

What Are We Actually Benchmarking in Robot Manipulation?

A robotics benchmark score measures success under one fixed evaluation setup, yet is routinely treated as evidence of general manipulation capability. We identify four failure modes, each of which weakens or invalidates a benchmark's role as a valid proxy for that capability: shortcut solvability, lack of statistical significance, creeping overfitting, and data-source dependence. We propose one diagnostic per failure mode. We audit LIBERO, CALVIN, SimplerEnv, RoboCasa, and RoboTwin 2.0 under these diagnostics. LIBERO and CALVIN fail multiple diagnostics. RoboCasa and RoboTwin 2.0 fail fewer, despite appearing far less often in recent progress claims. On LIBERO, a 0.09B probe with no language encoder scores at or near reported SOTA, and most reported gains are not provably statistically significant. On CALVIN, randomizing block poses within the training range drops performance for every tested policy. We release the four diagnostics with reference implementations for authors and reviewers to apply before treating a benchmark score as evidence of progress. Code and artifacts are available at https://ripl.github.io/manipulation_benchmark_audit/.

Tianchong Jiang, Xiangshan Tan, Samuel Wheeler +3

Toyota Technological Institute at Chicago, Chicago, IL, USA · Argonne National Laboratory, Lemont, IL, USA · University of Chicago, Chicago, IL, USA

Benchmark Design Benchmark Contamination Robot Manipulation Benchmarks Benchmark Validity