_target_: text_recognizer.data.tokenizer.Tokenizer
extra_symbols: ["\n"]