tokenizer.fit_on_texts() https://stackoverflow.com/questions/65238940/how-can-i-use-fit-with-generator-on-seq2seq-model