rotobart
rotobart copied to clipboard
DataCollatorForDenoisingTasks
First, thank you for your public code. But I have a questeion, In "DataCollatorForDenoisingTasks" code, I find batch["decoder_input_ids"] = self.shift_tokens_right(batch["input_ids"]) is original order data, but the batch["labels"] = self.add_whole_word_mask(batch["input_ids"], do_permutate) is permute data, So I'm very confused about this, I think the labels should be original order data.