论文标题
语言模型是多语言链的推理者
Language Models are Multilingual Chain-of-Thought Reasoners
论文作者
论文摘要
我们评估了多语言设置中大语言模型的推理能力。我们通过手动翻译250个级别的数学问题，从GSM8K数据集（Cobbe等，2021）中介绍了多种多样的语言，从而介绍了多语言小学数学（MGSM）基准。我们发现，通过增加的模型量表来解决MGSM问题的能力，即使在孟加拉语和斯瓦希里语等代表性不足的语言中，模型也具有强大的多语言推理能力。最后，我们表明语言模型的多语言推理能力扩展到其他任务，例如常识性推理和语言语义判断。 MGSM基准可以在https://github.com/google-research/url-nlp上公开获得。
We evaluate the reasoning abilities of large language models in multilingual settings. We introduce the Multilingual Grade School Math (MGSM) benchmark, by manually translating 250 grade-school math problems from the GSM8K dataset (Cobbe et al., 2021) into ten typologically diverse languages. We find that the ability to solve MGSM problems via chain-of-thought prompting emerges with increasing model scale, and that models have strikingly strong multilingual reasoning abilities, even in underrepresented languages such as Bengali and Swahili. Finally, we show that the multilingual reasoning abilities of language models extend to other tasks such as commonsense reasoning and word-in-context semantic judgment. The MGSM benchmark is publicly available at https://github.com/google-research/url-nlp.
