FLORES-200: Machine Translation Benchmark for 200 Languages

FLORES-200 is a benchmark dataset for machine translation between English and low-resource languages. It was created by Facebook and doubles the language coverage of the earlier FLORES-101 dataset. The dataset is managed by the Open Language Data Initiative, with a newer version available as 'flores_plus'.

Use Cases

• Benchmarking machine translation model performance based on the described multilingual parallel text.
• Training translation models for low-resource languages based on the dataset's stated focus.
• Evaluating cross-lingual transfer learning techniques using the expanded language coverage mentioned in the description.

Strengths

• Covers 200 languages, doubling the coverage of its predecessor FLORES-101.
• Explicitly designed for benchmarking machine translation, providing a clear evaluation purpose.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Column-level documentation is absent; field semantics must be inferred after download.
• Last updated 2026-05-29 08:38:45; freshness should be verified.

Provenance

Source

facebook

Freshness

Last updated 2026-05-29 08:38:45.