Saved in:
Bibliographic Details
Main Authors: Gao, Zitian, Chen, Lynx, Luo, Haoming, Zhou, Joey, Dai, Bryan
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2505.20282
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913998995193856
author Gao, Zitian
Chen, Lynx
Luo, Haoming
Zhou, Joey
Dai, Bryan
author_facet Gao, Zitian
Chen, Lynx
Luo, Haoming
Zhou, Joey
Dai, Bryan
contents We trained 13,440 large language models and found that entropy minimization requires only a single unlabeled data and 10 steps optimization to achieve performance improvements comparable to or even greater than those obtained using thousands of data and carefully designed rewards in rule-based reinforcement learning. This striking result may prompt a rethinking of post-training paradigms for large language models. Our code is avaliable at https://github.com/zitian-gao/one-shot-em.
format Preprint
id arxiv_https___arxiv_org_abs_2505_20282
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle One-shot Entropy Minimization
Gao, Zitian
Chen, Lynx
Luo, Haoming
Zhou, Joey
Dai, Bryan
Computation and Language
We trained 13,440 large language models and found that entropy minimization requires only a single unlabeled data and 10 steps optimization to achieve performance improvements comparable to or even greater than those obtained using thousands of data and carefully designed rewards in rule-based reinforcement learning. This striking result may prompt a rethinking of post-training paradigms for large language models. Our code is avaliable at https://github.com/zitian-gao/one-shot-em.
title One-shot Entropy Minimization
topic Computation and Language
url https://arxiv.org/abs/2505.20282