This tokenizer seems to be the same as what was used to train the model.
· Sign up or log in to comment