探究大型多语言语言模型的翻译性能：以BLOOM为例
Investigating the Translation Performance of a Large Multilingual Language Model: the Case of BLOOM
AI总结：
本研究评估了多语言模型BLOOM的翻译性能，发现few-shot设置显著改善0-shot的过度生成和语言错误，并探讨了提示设计、模型规模等因素的影响。
摘要：
自然语言处理社区最近发布了新的大型开放获取多语言语言模型BLOOM（BigScience等，2022），覆盖46种语言。我们通过评估其在多个数据集（WMT、Flores-101和DiaBLa）和语言对（高资源和低资源）上的机器翻译性能，聚焦于BLOOM的多语言能力。我们的结果表明，0-shot性能存在过度生成和以错误语言生成的问题，但在few-shot设置中这一问题得到极大改善，对许多语言对取得了非常好的结果。我们研究了多个方面，包括提示设计、模型规模、跨语言迁移以及话语语境的使用。
英文摘要：
The NLP community recently saw the release of a new large open-access multilingual language model, BLOOM (BigScience et al., 2022) covering 46 languages. We focus on BLOOM's multilingual ability by evaluating its machine translation performance across several datasets (WMT, Flores-101 and DiaBLa) and language pairs (high- and low-resourced). Our results show that 0-shot performance suffers from overgeneration and generating in the wrong language, but this is greatly improved in the few-shot setting, with very good results for a number of language pairs. We study several aspects including prompt design, model sizes, cross-lingual transfer and the use of discursive context.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
