Fine-tuning large language models for multilingual downstream tasks requires
a diverse set of languages to capture the nuances and structures of different
linguistic contexts effectively. While the specific number varies depending on
the desired scope and target languages, we argue that the number of languages,
language exposure, and similarity that incorporate the selection of languages
for fine-tuning are some important aspects to examine. By fine-tuning large
multilingual models on 1 to 52 languages, this paper answers one question: How
many languages are needed in instruction fine-tuning for multilingual tasks? We
investigate how multilingual instruction fine-tuned models behave on
multilingual benchmarks with an increasing number of languages and discuss our
findings from the perspective of language exposure and similarity.

这篇论文研究了多语言下游任务的大型语言模型的微调，探讨了在微调中需要多少种语言以及语言曝光度和相似度对模型性能的影响。

Lucky 52: 需要多少语言来指导大规模语言模型的微调？

Lucky 52: How Many Languages Are Needed to Instruction Fine-Tune Large  Language Models?

English-based Vision-Language Pre-training (VLP) has achieved great success
in various downstream tasks. Some efforts have been taken to generalize this
success to non-English languages through Multilingual Vision-Language
Pre-training (M-VLP). However, due to the large number of languages, M-VLP
models often require huge computing resources and cannot be flexibly extended
to new languages. In this work, we propose a \textbf{M}ulti\textbf{L}ingual
\textbf{A}cquisition (MLA) framework that can easily generalize a monolingual
Vision-Language Pre-training model into multilingual. Specifically, we design a
lightweight language acquisition encoder based on state-of-the-art monolingual
VLP models. We further propose a two-stage training strategy to optimize the
language acquisition encoder, namely the Native Language Transfer stage and the
Language Exposure stage. With much less multilingual training data and
computing resources, our model achieves state-of-the-art performance on
multilingual image-text and video-text retrieval benchmarks.

本研究提出了一种 MultiLingual Acquisition（MLA）框架，该框架可以将单语言视觉语言预训练模型轻松推广到多语言环境中，并采用两阶段训练策略来优化语言获取编码器。通过更少的数据和计算资源，我们的模型在多语言图像文本和视频文本检索基准上实现了最先进的性能。