Fix initialization in hubert_pretrain_model (#2846)

Summary:
address https://github.com/pytorch/audio/issues/2845

Pull Request resolved: https://github.com/pytorch/audio/pull/2846

Reviewed By: carolineechen

Differential Revision: D41251624

Pulled By: nateanl

fbshipit-source-id: 1a363d2314d6a452f35c109b9730da64ada5a2fd

torchaudio/models/wav2vec2/components.py

@@ -197,6 +197,7 @@ class ConvolutionalPositionalEmbedding(Module):
     ):
         super().__init__()
         self.embed_dim = embed_dim
+        self.kernel_size = kernel_size
         self.conv = nn.Conv1d(
             in_channels=embed_dim,
             out_channels=embed_dim,