Datasets
Every dataset is content-addressed. Downloads produce a signed access record.
eu-legal-corpusv3.2.0
12.4M rows · 42.1 GB · CC-BY 4.0
medical-notes-anonv1.8.4
4.1M rows · 18.9 GB · Restricted
swe-bench-plusv0.9.1
2,904 rows · 3.2 GB · MIT
voice-multilingual-24v2.0.0
980K clips rows · 1.1 TB · CC-BY-NC