What it does
UDPipe provides a language-processing pipeline for tokenization, grammatical tagging, lemmatization and dependency parsing.
For a local UDPipe 1 job, choose the trained model, UTF-8 input and tokenization, tagging or parsing stages, then send the structured result to standard output or a named file. Output templates and formats have their own controls. The manual says the default loads the whole input before processing; immediate mode processes complete sentence blocks sooner. UDPipe 2 is a separate Python prototype, and both versions are available through a hosted service. The engine's MPL licence does not remove non-commercial or additional conditions attached to supplied language models.
Strengths
- A complete tokenization-to-parsing pipeline
- Standalone application and service interfaces
Best for
- Preparing a corpus for grammatical analysis
Sources
- UDPipe 2 Versions | ÚFAL (Oct 26, 2026)
- UDPipe 1 | ÚFAL (Oct 26, 2026)
- UDPipe Versions | ÚFAL (Oct 26, 2026)
- UDPipe 1 | ÚFAL (Oct 26, 2026)
- UDPipe 1 User's Manual (Oct 26, 2026)