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 0150d818c..30d8d9b78 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 708eecfc4..cbfc809e9 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
@@ -493,7 +493,7 @@ class Data2VecAudioModel(BaseFairseqModel):
             #features = torch.cat([features, prompt])
             features = torch.cat([prompt, features])
             prompt_padding_mask = torch.zeros(promt.size()).type(torch.BoolTensor).to(features.device)
-            padding_mask
+            padding_mask = torch.cat([prompt_padding_mask, padding_mask])
 
         features = self.layer_norm(features)