Pretrained multilingual large language models have typically used heuristic temperature-based sampling to balance between different languages. However previous work has not systematically evaluated the efficacy of different pretraining language distributions across model scales. In this paper, we propose a new sampling method, UniMax, that delivers more uniform coverage of head languages while mitigating overfitting on tail languages by explicitly capping the number of repeats over each language's corpus. We perform an extensive series of ablations testing a range of sampling strategies on a suite of multilingual benchmarks, while varying model scale. We find that UniMax outperforms standard temperature-based sampling, and the benefits persist as scale increases. As part of our contribution, we release: (i) an improved and refreshed mC4 multilingual corpus consisting of 29 trillion characters across 107 languages, and (ii) a suite of pretrained umT5 model checkpoints trained with UniMax sampling.

翻译：预训练的多语言大型语言模型通常采用启发式的基于温度的采样技术来平衡不同的语言。然而，之前的研究并没有系统地评估不同预训练语言分布在不同模型规模下的效果。在本文中，我们提出了一种新的采样方法——UniMax，它在明确限制每个语言语料库中的重复次数的同时提供了更均匀的对头部语言的覆盖，并减轻了对尾部语言的过度拟合问题。我们在一系列多语言基准测试上进行了广泛的消融实验，同时变化模型规模测试不同的采样策略。我们发现UniMax优于标准温度采样，而且随着规模的增加，这些优势依然存在。作为我们的贡献的一部分，我们发布了：(i)一个包含107种语言、总字符数达29万亿字符的改进和刷新的mC4多语言语料库，(ii)采用UniMax采样训练的预训练umT5模型的检查点套件。