metabench - Paper Data

2024/07/04 by Alex Kipnis, Kipnis, Alex, Konstantinos Voudouris +5 · 6 citations
Computer Science · #Computation and Language (cs.CL) #FOS: Computer and information sciences #Machine Learning (cs.LG) #Machine Learning (stat.ML) #Natural Language Processing Techniques #Text Readability and Simplification #Topic Modeling

paper · pdf · doi:10.48550/arxiv.2407.12844

openalex publication_date 2024/07/04 · openalex created_date 2025/01/04 · openalex updated_date 2026/07/28

Abstract

Item-wise accuracies in six benchmarks from Open LLM Leaderboard 1 scraped from huggingface.co and used for metabench analyses and construction. Datasets with RMSE's for random benchmark subsets are used as reference in the paper and are included here.

Cited by

• Scales++: Compute Efficient Evaluation Subset Selection with Cognitive Scales Embeddings
• Adaptive Testing for LLM Evaluation: A Psychometric Alternative to Static Benchmarks
• Rethinking LLM Evaluation: Can We Evaluate LLMs with 200x Less Data?
• DISCO: Diversifying Sample Condensation for Efficient Model Evaluation
• Predicting LLM Reasoning Performance with Small Proxy Model
• Fluid Language Model Benchmarking

Related

• Mapping Overlaps in Benchmarks through Perplexity in the Wild
• MR-GSM8K: A Meta-Reasoning Benchmark for Large Language Model Evaluation
• Benchmark Profiling: Mechanistic Diagnosis of LLM Benchmarks
• Line Goes Up? Inherent Limitations of Benchmarks for Evaluating Large Language Models
• Forget What You Know about LLMs Evaluations -- LLMs are Like a Chameleon