`add_prefix_space=True` option for the BPE tokenizer

🤗Transformers
1

Hello,
I understand that when I add the add_prefix_space=True option in the BPE tokenizer statement, the tokenizer will add a space in the beginning of every sequence.

Is there some specific advantages of using the add_prefix_space=True option for BPE tokenizer (compared to when I don’t use the option)? All my sequences start without a space in the beginning.

Thanks,