The FLORES-101 Evaluation Benchmark for Low-Resource and Multilingual Machine Translation

TLDR

Low‑resource and multilingual machine translation lacks robust evaluation benchmarks, and existing ones suffer from limited language coverage, domain restriction, or low quality due to semi‑automatic construction. The authors introduce FLORES‑101, a high‑quality, high‑coverage benchmark of 3,001 English Wikipedia sentences translated into 101 languages, to advance low‑resource and multilingual MT research. The benchmark comprises 3,001 English Wikipedia sentences translated into 101 languages by professional translators via a carefully controlled process. FLORES‑101 enables more accurate evaluation of model performance on low‑resource languages and many‑to‑many multilingual systems thanks to its multilingual alignment.

Abstract

One of the biggest challenges hindering progress in low-resource and multilingual machine translation is the lack of good evaluation benchmarks. Current evaluation benchmarks either lack good coverage of low-resource languages, consider only restricted domains, or are low quality because they are constructed using semi-automatic procedures. In this work, we introduce the FLORES-101 evaluation benchmark, consisting of 3001 sentences extracted from English Wikipedia and covering a variety of different topics and domains. These sentences have been translated in 101 languages by professional translators through a carefully controlled process. The resulting dataset enables better assessment of model quality on the long tail of low-resource languages, including the evaluation of many-to-many multilingual translation systems, as all translations are multilingually aligned. By publicly releasing such a high-quality and high-coverage dataset, we hope to foster progress in the machine translation community and beyond.