BART Fine-Tuning Resources/Help

hello. i’m in need of some assistance. i’m trying to fine-tune a bart (not bert) model using huggingface transformers, but i can’t find what the input and output dataset key names are for it anywhere. the keys aren’t ‘input’ and ‘labels’. there aren’t many helpful resources i could find when it comes to learning how to fine-tune bart specifically. is there a method in the model, tokenizer, or trainer that can give me this info? any help would be greatly appreciated.