多领域机器翻译基准数据集
收藏资源简介:
多领域机器翻译基准数据集由上海交通大学、同壹实验室和NLP2CT实验室(澳门大学)共同创建,旨在评估大型语言模型在多领域机器翻译中的表现。该数据集包含25个德英和22个中英测试集,涵盖新闻、医疗、法律、IT等15个领域,总条数为47。数据集的创建过程结合了OPUS、WMT、TedTalks等多个来源的数据,并通过精细的领域分类和标注确保数据的多样性和代表性。该数据集主要应用于机器翻译模型的多领域适应性和泛化能力评估,旨在解决现有模型在不同领域翻译质量不一致的问题。
The Multi-Domain Machine Translation Benchmark Dataset was jointly developed by Shanghai Jiao Tong University, Tongyi Lab, and NLP2CT Lab (University of Macau). It is designed to evaluate the performance of large language models (LLMs) in multi-domain machine translation. This dataset comprises 25 German-English and 22 Chinese-English test sets, covering 15 domains including news, medical, legal, IT and other fields, with a total of 47 translation sample pairs. The dataset was constructed using data from multiple sources such as OPUS, WMT, and TedTalks, and ensures data diversity and representativeness through precise domain classification and annotation. This dataset is mainly applied to the evaluation of multi-domain adaptability and generalization ability of machine translation models, with the goal of resolving the problem of inconsistent translation quality of existing models across different domains.

- 1Large Language Model for Multi-Domain Translation: Benchmarking and Domain CoT Fine-tuning上海交通大学、同壹实验室、NLP2CT实验室(澳门大学) · 2024年



