Skip to content

Puhti and Mahti computing services have been decommissioned. Puhti and Mahti login nodes and storage services will remain available until 15 October 2026, but are no longer covered by service contracts. Please clean up and migrate your data to Roihu ASAP. See Roihu data migration guide for instructions.

UDPipe

UDPipe is a trainable pipeline for tokenization, tagging, lemmatization and dependency parsing of natural language text. It is developed at the Institute of Formal and Applied Linguistics (ÚFAL), Charles University, and works with pre-trained models for the Universal Dependencies treebanks, covering a wide range of languages.

Available

  • Puhti: 1.4.0
  • Roihu: 1.4.0

License

The UDPipe software is licensed under the Mozilla Public License 2.0. The pre-trained language models are distributed under the CC BY-NC-SA license and are therefore restricted to non-commercial use; some models may carry additional conditions from their original training data.

Usage

Initialize UDPipe with:

module load udpipe

The pre-trained models are available in the $UD_MODELS directory. List the installed models with:

ls $UD_MODELS

To tag and parse text, give UDPipe a model file and the input. For example, to process Finnish text given one sentence per line (horizontal input):

echo "Hän istuu." | udpipe --tag --parse --input=horizontal --immediate \
  $UD_MODELS/finnish-ftb-ud-2.5-191206.udpipe

More information