[Corpora-List] LIMA 3.0 beta : 60+ languages with Deep Learning models

Gaël de Chalendar gael.de-chalendar at cea.fr
Tue Feb 11 12:37:52 CET 2020


We are pleased to present the first beta of the new major version of LIMA, the LIbre Multilingual Analyzer, with state of the art performance and now fully multilingual with more than 60 language models available.

It was possible thanks to the Universal Dependencies [1] initiative offering training corpora for numerous languages and to new deep learning-based neural models.

The LIMA features (configurability, pipelines…) making it very versatile and quick to adapt to new domains remain available with this version. In particular, the powerful Modex (Extraction Modules) feature can still be used in the same way.

Note that this is a beta version. Please note that there is some known problems (See release notes [2]).

Note also that LIMA 3.0 beta is available for Ubuntu 18.04 only. To test it, please refer to [2]

[1] https://universaldependencies.org/ [2] https://github.com/aymara/lima/wiki/DeepLima-beta

-- Gael de Chalendar CEA LIST Laboratoire d'Analyse Sémantique Texte et Image (Text and Image Semantic Analysis Laboratory)

CEA Saclay Nano-INNOV DRT/LIST/DIASI/SIALV BAT 861 PC 184 F-91191 Gif-sur-Yvette Cedex

Tél.:+33.1.69.08.01.50 Fax:+33.1.69.08.01.15 Email : Gael.D.O.T.de-Chalendar.A at T.cea.D.O.T.fr -------------- next part -------------- A non-text attachment was scrubbed... Name: smime.p7s Type: application/pkcs7-signature Size: 3493 bytes Desc: not available URL: <https://mailman.uib.no/public/corpora/attachments/20200211/5f50e282/attachment.p7s>



More information about the Corpora mailing list