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 903071da0..288955dc0 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/.prompt_tuning.py.swp b/egs/librispeech/ASR/pruned_transducer_stateless_d2v_v2/.prompt_tuning.py.swp
index 8c057dcbe..a0cce3133 100644
Binary files a/egs/librispeech/ASR/pruned_transducer_stateless_d2v_v2/.prompt_tuning.py.swp and b/egs/librispeech/ASR/pruned_transducer_stateless_d2v_v2/.prompt_tuning.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 cbfc809e9..d2513b298 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
@@ -491,6 +491,7 @@ class Data2VecAudioModel(BaseFairseqModel):
         ## for prompt tuning
         if prompt is not None:
             #features = torch.cat([features, prompt])
+            prompt = prompt.expand((features.size()[0], prompt.size()[0], prompt.size()[1])
             features = torch.cat([prompt, features])
             prompt_padding_mask = torch.zeros(promt.size()).type(torch.BoolTensor).to(features.device)
             padding_mask = torch.cat([prompt_padding_mask, padding_mask])