dbmdz/bert-base-german-cased

In this repository the MDZ Digital Library team (dbmdz) at the Bavarian State.Library open sources another German BERT model . The source data for the model consists of a recent Wikipedia dump, EU Bookshop corpus, CommonCrawl, ParaCrawl and News Crawl . The model is trained with an initial.sequence length of 512 subwords and was performed for 1.5M steps . This release includes both cased and uncased models . All models are available on the Huggingface model hub. Currently only PyTorch-Transformers-compatible weights are available . For questions about our BERT models just open an issue with just an issue here, please,