prajjwal1

/

bert-tiny

like 103

These models are ported from Google repo BERT checkpoint. So you should refer to the Google BERT repository for training code and tokenizer info.