1,000 rows from random languages and splits of xP3x for each dataset represented in [xP3x](https://huggingface.co/datasets/Muennighoff/xP3x).