Intended usage of num_special_tokens?

[LLYX]

From what I understand, these are supposed to be reserved for oov values. Is the intended usage to set oov values in the input to some negative number and overwrite the offset? That is what it seems like it would take to achieve the desired outcome, but also seems somewhat confusing and clunky to do. Or perhaps I am misunderstanding its purpose? Thanks!

[lucidrains]

@LLYX it was originally designed for self-supervised learning with electra, for adding masking token etc

[lucidrains]

@LLYX i should probably just remove it though, since that line of research never took off