A version of Google’s BERT deep transfer learning model for Finnish, developed by the TurkuNLP Group. The model can be fine-tuned to achieve state-of-the-art results for various Finnish natural language processing tasks.
FinBERT has been pre-trained for 1 million steps on over 3 billion tokens (24B characters) of Finnish text drawn from news, online discussion, and internet crawls.
For more information see the FinBERT’s project page
FinBERT Kielipankki version: Kielipankki offers a version of Google’s BERT deep transfer learning model for Finnish. It is installed in CSC’s Puhti cluster and can be used via the pytorch 1.4 module. For details see /appl/data/kielipankki/bert_models/README.txt
This resource group page has a Persistent Identifier: http://urn.fi/urn:nbn:fi:lb-2021110401
