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 4139bcfe4..3a15d829d 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 c2960c8c3..ab47fbdf7 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
@@ -42,7 +42,7 @@ class TransformerEncoderAdapter(TransformerEncoder):
         self.adapters = ResidualAdapterModule(proj_dim=512)
 
         for p in self.adapters.parameters():
-            p.data /= 10.
+            p.data /= 5.
         #p.data = nn.Parameter(torch.zeros(p.size()).to('cuda'))
         #p.data = nn.Parameter(torch.randn(p.size()).to('cuda')/20.)