diff --git a/docs/source/model_summary.mdx b/docs/source/model_summary.mdx
index e77550ab9a7..6123bc2bcac 100644
--- a/docs/source/model_summary.mdx
+++ b/docs/source/model_summary.mdx
@@ -542,6 +542,10 @@ As mentioned before, these models keep both the encoder and the decoder of the o
+
+
+
+
[BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension](https://arxiv.org/abs/1910.13461), Mike Lewis et al.