Salvato in:
Dettagli Bibliografici
Autori principali: Saxon, Michael, Holtzman, Ari, West, Peter, Wang, William Yang, Saphra, Naomi
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:https://arxiv.org/abs/2407.16711
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866911971376365568
author Saxon, Michael
Holtzman, Ari
West, Peter
Wang, William Yang
Saphra, Naomi
author_facet Saxon, Michael
Holtzman, Ari
West, Peter
Wang, William Yang
Saphra, Naomi
contents Modern language models (LMs) pose a new challenge in capability assessment. Static benchmarks inevitably saturate without providing confidence in the deployment tolerances of LM-based systems, but developers nonetheless claim that their models have generalized traits such as reasoning or open-domain language understanding based on these flawed metrics. The science and practice of LMs requires a new approach to benchmarking which measures specific capabilities with dynamic assessments. To be confident in our metrics, we need a new discipline of model metrology -- one which focuses on how to generate benchmarks that predict performance under deployment. Motivated by our evaluation criteria, we outline how building a community of model metrology practitioners -- one focused on building tools and studying how to measure system capabilities -- is the best way to meet these needs to and add clarity to the AI discussion.
format Preprint
id arxiv_https___arxiv_org_abs_2407_16711
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Benchmarks as Microscopes: A Call for Model Metrology
Saxon, Michael
Holtzman, Ari
West, Peter
Wang, William Yang
Saphra, Naomi
Software Engineering
Computation and Language
Modern language models (LMs) pose a new challenge in capability assessment. Static benchmarks inevitably saturate without providing confidence in the deployment tolerances of LM-based systems, but developers nonetheless claim that their models have generalized traits such as reasoning or open-domain language understanding based on these flawed metrics. The science and practice of LMs requires a new approach to benchmarking which measures specific capabilities with dynamic assessments. To be confident in our metrics, we need a new discipline of model metrology -- one which focuses on how to generate benchmarks that predict performance under deployment. Motivated by our evaluation criteria, we outline how building a community of model metrology practitioners -- one focused on building tools and studying how to measure system capabilities -- is the best way to meet these needs to and add clarity to the AI discussion.
title Benchmarks as Microscopes: A Call for Model Metrology
topic Software Engineering
Computation and Language
url https://arxiv.org/abs/2407.16711