bio-lm
bio-lm copied to clipboard
Token IDs for <mask> and <unk> are the same for RoBERTa-base-PM-Voc-hf
I noticed that the IDs for both are 3, while for roberta-base
the ID for <mask>
is 50264 and 3 for <unk>
. Is this a bug or is it supposed to be like this?