RepBench: Compiling Benchmarks into Capability Representations for Large Language Models

• URL: http://arxiv.org/abs/2607.28008v1
• Date: Thu, 30 Jul 2026 10:56:44 GMT
• Title: RepBench: Compiling Benchmarks into Capability Representations for Large Language Models
• Abstract summary: RepBench is a benchmark-grounded data layer for capability-aligned representation probing.<n>We crawl 13,427 benchmark papers yields a taxonomy of 182 capability clusters in 13 families; harvesting 353 public benchmark datasets yields 46,149 audited probe texts covering 94 capabilities.
• Score: 5.276992083399684
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Representation engineering reads and steers capability directions in large language models, yet methods are typically evaluated on paper-specific synthetic data. The resulting measurements are difficult to compare or reproduce and may reflect surface patterns rather than capabilities. We present RepBench, a benchmark-grounded data layer for capability-aligned representation probing. Crawling 13,427 benchmark papers yields a taxonomy of 182 capability clusters in 13 families; harvesting 353 public benchmark datasets yields 46,149 audited probe texts covering 94 capabilities, each supported by at least two independent benchmarks. This multi-benchmark design reduces dependence on any single source: raw per-text vectors exhibit no natural cluster granularity, whereas benchmark-pooled capability vectors show an interior clustering optimum at a small number of clusters on all 12 evaluated models, with low agreement to the human taxonomy. Under cross-benchmark transfer evaluation across twelve models completed by all four readouts, difference-in-means attains the highest model-level mean on ten models, while logistic regression wins the most capability-model cells. This disagreement shows that the readout method and aggregation criterion are meaningful evaluation dimensions. The pipeline, corpus, and evaluation code are released as a reusable closed-loop workflow.

Related papers

• PhenoBench: Mapping What a Deeply Phenotyped Human Cohort Can Tell Us [0.7084863262134483]
  We present PhenoBench, an executable benchmark built around the Human Phenotype Project.<n>The benchmark defines 90 clinically grounded tasks across 15 domains and 26 input modalities.<n> Measurements showed question- and representation-dependent predictive value, including positive, near-zero, and negative changes in held-out performance.
  arXiv  Detail & Related papers  (2026-09-05T13:15:40Z)
• A Statistical Audit of Physical AI Benchmark Redundancy [0.0]
  Physical AI models are evaluated on suites of benchmarks that differ across model reports, leaving the model-by-benchmark matrix sparse.<n>We construct a matrix of 51 models on 12 physical AI benchmarks, selected from a registry of 51 benchmarks and 152 models by reporting density.<n>We measure how much information the benchmarks share and show quantitative evidence of Redundancy.
  arXiv  Detail & Related papers  (2026-08-26T15:54:15Z)
• PACE: A Proxy for Agentic Capability Evaluation [60.3743414796937]
  We introduce PACE, a framework that constructs proxy benchmarks by selecting instances from existing non-agentic evaluations.<n>Experiments across 14 models, 4 agentic benchmarks, and 19 non-agentic benchmarks show that PACE-Bench predicts agentic scores with leave-one-out cross-validation (LOOCV) mean absolute error (MAE) under 4%, Spearman correlation above 0.80, and pairwise model-ranking accuracy around 85%, all at much less than 1% of the full agentic evaluation cost.
  arXiv  Detail & Related papers  (2026-07-02T10:59:03Z)
• An Embarrassingly Simple Graph Heuristic Reveals Shortcut-Solvable Benchmarks for Sequential Recommendation [50.09718257952108]
  Sequentialsolv has increasingly shifted toward generative recommenders that combine sequential patterns with semantic item information.<n>These methods are often evaluated on a small set of widely used benchmarks, raising a key question: do these benchmarks actually require the advanced modeling capabilities that modern generative recommenders claim to provide?<n>We conduct a benchmark audit with an intentionally simple graph. Starting from only the last one or two items, it retrieves candidates from a few-hop item-transition graph and ranks them by item-feature similarity.
  arXiv  Detail & Related papers  (2026-05-08T02:00:11Z)
• GIANTS: Generative Insight Anticipation from Scientific Literature [84.95947892931142]
  We introduce insight anticipation, a generation task in which a model predicts a downstream paper's core insight from its foundational parent papers.<n>We evaluate models using an LM judge that scores similarity between generated and ground-truth insights, and show that these similarity scores correlate with expert human ratings.<n>We present GIANTS-4B, an LM trained via reinforcement learning (RL) to optimize insight anticipation using these similarity scores as a proxy reward.
  arXiv  Detail & Related papers  (2026-04-10T18:13:55Z)
• BenchBench: Benchmarking Automated Benchmark Generation [10.44497524694021]
  BenchBench is a pipeline and dataset for benchmarking automated benchmark generation.<n>We generate 16.7K items, retain 15K core items post-filtering, and produce 152K graded model--item responses.
  arXiv  Detail & Related papers  (2026-03-21T13:05:32Z)
• Learning More from Less: Unlocking Internal Representations for Benchmark Compression [37.69575776639016]
  We introduce REPCORE, which aligns heterogeneous hidden states into a unified latent space to construct representative coresets.<n>Experiments on five benchmarks and over 200 models show consistent gains over output-based baselines in ranking correlation and estimation accuracy.
  arXiv  Detail & Related papers  (2026-01-31T13:11:39Z)
• Benchmark^2: Systematic Evaluation of LLM Benchmarks [66.2731798872668]
  We propose Benchmark2, a comprehensive framework comprising three complementary metrics.<n>We conduct experiments across 15 benchmarks spanning mathematics, reasoning, and knowledge domains.<n>Our analysis reveals significant quality variations among existing benchmarks and demonstrates that selective benchmark construction can achieve comparable evaluation performance.
  arXiv  Detail & Related papers  (2026-01-07T14:59:03Z)
• Beyond One-Size-Fits-All: Tailored Benchmarks for Efficient Evaluation [19.673388630963807]
  We present TailoredBench, a method that conducts customized evaluation tailored to each target model.<n>A Global-coreset is first constructed as a probe to identify the most consistent source models for each target model.<n>A scalable K-Medoids clustering algorithm is proposed to extend the Global-coreset to a tailored Native-coreset for each target model.
  arXiv  Detail & Related papers  (2025-02-19T09:31:50Z)
• CLOVER: A Test Case Generation Benchmark with Coverage, Long-Context, and Verification [71.34070740261072]
  This paper presents a benchmark, CLOVER, to evaluate models' capabilities in generating and completing test cases.<n>The benchmark is containerized for code execution across tasks, and we will release the code, data, and construction methodologies.
  arXiv  Detail & Related papers  (2025-02-12T21:42:56Z)
• Anchor Points: Benchmarking Models with Much Fewer Examples [88.02417913161356]
  In six popular language classification benchmarks, model confidence in the correct class on many pairs of points is strongly correlated across models. We propose Anchor Point Selection, a technique to select small subsets of datasets that capture model behavior across the entire dataset. Just several anchor points can be used to estimate model per-class predictions on all other points in a dataset with low mean absolute error.
  arXiv  Detail & Related papers  (2023-09-14T17:45:51Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.