This is a large DeBERTa V2 model pre-trained on Japanese text using character-level whole word masking. It is designed for Japanese natural language processing tasks including fill-mask, sequence classification, and token classification. The model is hosted on Hugging Face and can be used with the Transformers library.
Deberta V2 Large Japanese Char Wwm is a Foundation models & chat product. It focuses on performing masked language modeling and downstream NLP tasks on Japanese text with character-aware modeling. It is built as an open-source project for developers. Deberta V2 Large Japanese Char Wwm is open source under the Apache-2.0 license. The product ships for the web and API.
Kyoto University NLP Lab builds and maintains Deberta V2 Large Japanese Char Wwm, and the product first shipped in 2017. Development happens publicly on GitHub with 12k stars and 172 commits in the last 90 days. Key capabilities include fill-Mask, Japanese NLP, and Character-level Tokenization.
Latest indexed changes and source events
ku-nlp/deberta-v2-large-japanese-char-wwm verified by the PulseGate indexer
