diff --git a/examples/seq2seq/finetune.py b/examples/seq2seq/finetune.py
index a095c318b12e..156b4695a67e 100755
--- a/examples/seq2seq/finetune.py
+++ b/examples/seq2seq/finetune.py
@@ -148,7 +148,7 @@ def _step(self, batch: dict) -> Tuple:
             self.save_readable_batch(batch)
 
         outputs = self(src_ids, attention_mask=src_mask, decoder_input_ids=decoder_input_ids, use_cache=False)
-        lm_logits = outputs[0]
+        lm_logits = outputs["logits"]
         if self.hparams.label_smoothing == 0:
             # Same behavior as modeling_bart.py, besides ignoring pad_token_id
             ce_loss_fct = torch.nn.CrossEntropyLoss(ignore_index=pad_token_id)