arXiv Open Access 2025

Enhancing Japanese Large Language Models with Reasoning Vectors

Carolina Minami Oguchi Leo Wei Koyo Kobayashi Hsin-Tai Wu Dipak Ghosal
Lihat Sumber

Abstrak

Post-training methods have improved the performance and enhanced the reasoning capability for mainstream large language models (LLMs), but the same is challenging for Japanese LLMs to achieve due to the amount of resources required. Inspired by task vectors that extract the change of weights before and after training, specifically for a certain task, we obtain reasoning vectors from reasoning LLMs and apply them to Japanese LLMs to boost their performance. While the resources available present a challenge to improve Japanese LLMs, we present a simple and effective way to obtain high improvement and hope to inspire for other languages.

Topik & Kata Kunci

Penulis (5)

C

Carolina Minami Oguchi

L

Leo Wei

K

Koyo Kobayashi

H

Hsin-Tai Wu

D

Dipak Ghosal

Format Sitasi

Oguchi, C.M., Wei, L., Kobayashi, K., Wu, H., Ghosal, D. (2025). Enhancing Japanese Large Language Models with Reasoning Vectors. https://arxiv.org/abs/2508.02913

Akses Cepat

Lihat di Sumber
Informasi Jurnal
Tahun Terbit
2025
Bahasa
en
Sumber Database
arXiv
Akses
Open Access ✓