wjy3326 opened this issue 4 years ago · 1 comments
What vocab file does the uniter model use? cased or uncased?
The default tokenizer in prepro.py is 'bert-base-cased'.