Problem fine-tuning a model with Seq2Seq Trainer

we need to see how you instantiate the model