ruSciBERT

Model was trained by Sber AI team and MLSA Lab of Institute for AI, MSU. If you use our model for your project, please tell us about it ( nikgerasimenko@gmail.com ).

Presentation at the AI Journey 2022

• Task: mask filling
• Type: encoder
• Tokenizer: bpe
• Dict size: 50265
• Num Parameters: 123 M
• Training Data Volume: 6.5 GB