Mismatching Tokens for Unicode Character

[FelixKirsch]

The unicode character "\uFFFD" is treated differently by the spacy tokenizer and the transformer tokenizer. Spacy creates a token, while transformer ignores the character. This leads to a mismatching in the tokens.