diff --git a/examples/speech_recognition/models/vggtransformer.py b/examples/speech_recognition/models/vggtransformer.py
index 97974360a4..bca0ae59a8 100644
--- a/examples/speech_recognition/models/vggtransformer.py
+++ b/examples/speech_recognition/models/vggtransformer.py
@@ -203,6 +203,7 @@ def prepare_transformer_decoder_params(
     relu_dropout,
 ):
     args = argparse.Namespace()
+    args.encoder_embed_dim = None
     args.decoder_embed_dim = input_dim
     args.decoder_attention_heads = num_heads
     args.attention_dropout = attention_dropout