EXAONE 3.5：面向真实应用场景的大语言模型系列
计算与语言
2026-01-05 v3
摘要
本技术报告介绍了由 LG AI Research 开发并发布的 EXAONE 3.5 指令微调语言模型。EXAONE 3.5 语言模型提供三种配置：32B、7.8B 和 2.4B。这些模型具备多项突出能力：1）在真实场景中展现出卓越的指令遵循能力，在七项基准测试中取得最高分数；2）具备出色的长上下文理解能力，在四项基准测试中达到顶尖性能；3）在九项通用基准测试中，与同等规模的最新开源模型相比具有竞争力。EXAONE 3.5 语言模型对所有人开放，可用于研究目的，并可从 https://huggingface.co/LGAI-EXAONE 下载。如需商业使用，请联系 LG AI Research 的官方联络点：[email protected]。
引用
@article{arxiv.2412.04862,
title = {EXAONE 3.5: Series of Large Language Models for Real-world Use Cases},
author = {Soyoung An and Kyunghoon Bae and Eunbi Choi and Kibong Choi and Stanley Jungkyu Choi and Seokhee Hong and Junwon Hwang and Hyojin Jeon and Gerrard Jeongwon Jo and Hyunjik Jo and Jiyeon Jung and Yountae Jung and Hyosang Kim and Joonkee Kim and Seonghwan Kim and Soyeon Kim and Sunkyoung Kim and Yireun Kim and Yongil Kim and Youchul Kim and Edward Hwayoung Lee and Haeju Lee and Honglak Lee and Jinsik Lee and Kyungmin Lee and Woohyung Lim and Sangha Park and Sooyoun Park and Yongmin Park and Sihoon Yang and Heuiyeen Yeen and Hyeongu Yun},
journal= {arXiv preprint arXiv:2412.04862},
year = {2026}
}