Pith. sign in

REVIEW 6 cited by

Efficient Benchmarking of Language Models

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2308.11696 v5 pith:LTLUJA5A submitted 2023-08-22 cs.CL cs.AIcs.CVcs.LG

Efficient Benchmarking of Language Models

classification cs.CL cs.AIcs.CVcs.LG
keywords benchmarkevaluationreliabilityefficientbenchmarkingbenchmarkscomputationcosts
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
Share X Bluesky LinkedIn Reddit HN
read the original abstract

The increasing versatility of language models (LMs) has given rise to a new class of benchmarks that comprehensively assess a broad range of capabilities. Such benchmarks are associated with massive computational costs, extending to thousands of GPU hours per model. However, the efficiency aspect of these evaluation efforts had raised little discussion in the literature. In this work, we present the problem of Efficient Benchmarking, namely, intelligently reducing the computation costs of LM evaluation without compromising reliability. Using the HELM benchmark as a test case, we investigate how different benchmark design choices affect the computation-reliability trade-off. We propose to evaluate the reliability of such decisions, by using a new measure -- Decision Impact on Reliability, DIoR for short. We find, for example, that a benchmark leader may change by merely removing a low-ranked model from the benchmark, and observe that a correct benchmark ranking can be obtained by considering only a fraction of the evaluation examples. Based on our findings, we outline a set of concrete recommendations for efficient benchmark design and utilization practices. To take a step further, we use our findings to propose an evaluation algorithm, that, when applied to the HELM benchmark, leads to dramatic cost savings with minimal loss of benchmark reliability, often reducing computation by x100 or more.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 6 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Holmes: A Benchmark to Assess the Linguistic Competence of Language Models

    cs.CL 2024-04 unverdicted novelty 7.0

    Holmes is a probing benchmark compiling over 200 datasets from 270 studies to evaluate linguistic competence across syntax, morphology, semantics, reasoning, and discourse in more than 50 language models.

  2. Consistent and Distinctive: LLM Benchmark Efficiency via Maximum Independent Set Prompt Selection on Similarity Graphs

    cs.CL 2026-05 unverdicted novelty 6.0

    A graph-based MIS prompt selection method on embedding similarity graphs yields reduced benchmark subsets with highly consistent LLM rankings (Kendall's W ≥ 0.90 in 99.2% of cases) and 25-48% size reduction at higher ...

  3. Welfare, Improvability, and Variance: A Principal-Agent Approach to Optimal Benchmark Item Aggregation

    cs.LG 2026-05 unverdicted novelty 6.0

    Models benchmarking as principal-agent game, derives welfare loss from welfare alignment, improvability and variance, and applies an audit framework to OLMES items.

  4. Query-efficient model evaluation using cached responses

    cs.LG 2026-05 unverdicted novelty 6.0

    DKPS-based methods leverage cached model responses to achieve equivalent benchmark prediction accuracy with substantially fewer queries than standard evaluation.

  5. Query-efficient model evaluation using cached responses

    cs.LG 2026-05 unverdicted novelty 6.0

    DKPS-based methods predict new model benchmark scores using cached responses, matching baseline mean absolute error with substantially fewer queries and an offline query selection approach.

  6. Quantum-Inspired Trace-Augmented Evidence Selection for Reasoning over Structured Hypothesis Spaces

    cs.AI 2026-06 unverdicted novelty 5.0

    EP-HUBO treats CoT evidence selection as higher-order unconstrained binary optimization over per-hypothesis pools with quality weights to improve aggregation on legal benchmarks.