diff --git a/egs/librispeech/ASR/pruned_transducer_stateless_d2v_v2/.data2vec_audio.py.swp b/egs/librispeech/ASR/pruned_transducer_stateless_d2v_v2/.data2vec_audio.py.swp
index e02802b80..66d3265fa 100644
Binary files a/egs/librispeech/ASR/pruned_transducer_stateless_d2v_v2/.data2vec_audio.py.swp and b/egs/librispeech/ASR/pruned_transducer_stateless_d2v_v2/.data2vec_audio.py.swp differ
diff --git a/egs/librispeech/ASR/pruned_transducer_stateless_d2v_v2/data2vec_audio.py b/egs/librispeech/ASR/pruned_transducer_stateless_d2v_v2/data2vec_audio.py
index 6e73be508..ad9222398 100644
--- a/egs/librispeech/ASR/pruned_transducer_stateless_d2v_v2/data2vec_audio.py
+++ b/egs/librispeech/ASR/pruned_transducer_stateless_d2v_v2/data2vec_audio.py
@@ -499,7 +499,7 @@ class Data2VecAudioModel(BaseFairseqModel):
             prompt_padding_mask = torch.zeros(prompt.size()[0], prompt.size()[1]).type(torch.BoolTensor).to(features.device)
             print(prompt_padding_mask.size())
             print(padding_mask.size())
-            padding_mask = torch.cat([prompt_padding_mask, padding_mask])
+            padding_mask = torch.cat([prompt_padding_mask, padding_mask], dim=1)
             print(padding_mask.size())
 
         features = self.layer_norm(features)