Détail du document
Identifiant
oai:arXiv.org:2407.17387
Sujet
Computer Science - Computation and...Auteur
Castricato, Louis Lile, Nathan Rafailov, Rafael Fränken, Jan-Philipp Finn, ChelseaCatégorie
Computer Science
Année
2024
Date de référencement
31/07/2024
Mots clés
diverse user persona datasetRésumé
The rapid advancement of language models (LMs) necessitates robust alignment with diverse user values.
However, current preference optimization approaches often fail to capture the plurality of user opinions, instead reinforcing majority viewpoints and marginalizing minority perspectives.
We introduce PERSONA, a reproducible test bed designed to evaluate and improve pluralistic alignment of LMs.
We procedurally generate diverse user profiles from US census data, resulting in 1,586 synthetic personas with varied demographic and idiosyncratic attributes.
We then generate a large-scale evaluation dataset containing 3,868 prompts and 317,200 feedback pairs obtained from our synthetic personas.
Leveraging this dataset, we systematically evaluate LM capabilities in role-playing diverse users, verified through human judges, and the establishment of both a benchmark, PERSONA Bench, for pluralistic alignment approaches as well as an extensive dataset to create new and future benchmarks.
The full dataset and benchmarks are available here: https://www.synthlabs.ai/research/persona.
Castricato, Louis,Lile, Nathan,Rafailov, Rafael,Fränken, Jan-Philipp,Finn, Chelsea, 2024, PERSONA: A Reproducible Testbed for Pluralistic Alignment
MELAS: Phenotype Classification into Classic-versus-Atypical Presentations
presentations mitochondrial strokelike patients variability phenotype clinical melas
