NoveltyBench: Evaluating Language Models for Humanlike Diversity

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zhang, Yiming, Diddee, Harshita, Holm, Susan, Liu, Hanchen, Liu, Xinyue, Samuel, Vinay, Wang, Barry, Ippolito, Daphne
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915436103204864
author Zhang, Yiming
Diddee, Harshita
Holm, Susan
Liu, Hanchen
Liu, Xinyue
Samuel, Vinay
Wang, Barry
Ippolito, Daphne
author_facet Zhang, Yiming
Diddee, Harshita
Holm, Susan
Liu, Hanchen
Liu, Xinyue
Samuel, Vinay
Wang, Barry
Ippolito, Daphne
contents Language models have demonstrated remarkable capabilities on standard benchmarks, yet they struggle increasingly from mode collapse, the inability to generate diverse and novel outputs. Our work introduces NoveltyBench, a benchmark specifically designed to evaluate the ability of language models to produce multiple distinct and high-quality outputs. NoveltyBench utilizes prompts curated to elicit diverse answers and filtered real-world user queries. Evaluating 20 leading language models, we find that current state-of-the-art systems generate significantly less diversity than human writers. Notably, larger models within a family often exhibit less diversity than their smaller counterparts, challenging the notion that capability on standard benchmarks translates directly to generative utility. While prompting strategies like in-context regeneration can elicit diversity, our findings highlight a fundamental lack of distributional diversity in current models, reducing their utility for users seeking varied responses and suggesting the need for new training and evaluation paradigms that prioritize diversity alongside quality.
format Preprint
id arxiv_https___arxiv_org_abs_2504_05228
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle NoveltyBench: Evaluating Language Models for Humanlike Diversity
Zhang, Yiming
Diddee, Harshita
Holm, Susan
Liu, Hanchen
Liu, Xinyue
Samuel, Vinay
Wang, Barry
Ippolito, Daphne
Computation and Language
Language models have demonstrated remarkable capabilities on standard benchmarks, yet they struggle increasingly from mode collapse, the inability to generate diverse and novel outputs. Our work introduces NoveltyBench, a benchmark specifically designed to evaluate the ability of language models to produce multiple distinct and high-quality outputs. NoveltyBench utilizes prompts curated to elicit diverse answers and filtered real-world user queries. Evaluating 20 leading language models, we find that current state-of-the-art systems generate significantly less diversity than human writers. Notably, larger models within a family often exhibit less diversity than their smaller counterparts, challenging the notion that capability on standard benchmarks translates directly to generative utility. While prompting strategies like in-context regeneration can elicit diversity, our findings highlight a fundamental lack of distributional diversity in current models, reducing their utility for users seeking varied responses and suggesting the need for new training and evaluation paradigms that prioritize diversity alongside quality.
title NoveltyBench: Evaluating Language Models for Humanlike Diversity
topic Computation and Language
url https://arxiv.org/abs/2504.05228