check codebook index range

egs/librispeech/ASR/vq_pruned_transducer_stateless2/hubert_code_indices.py

@@ -91,7 +91,11 @@ def compute_codeindices(
         )
 
         # [N, T, C]
-        codebook_indices = codebook_indices.to("cpu").numpy().astype(np.int8)
+        codebook_indices = codebook_indices.to("cpu").numpy()
+        assert np.all(
+            codebook_indices[np.where(codebook_indices < 0)] == -100
+        )
+        assert np.max(codebook_indices) < 256
 
         supervisions = batch["supervisions"]
         cut_list = supervisions["cut"]

egs/librispeech/ASR/vq_pruned_transducer_stateless2/model.py

@@ -133,6 +133,10 @@ class Transducer(nn.Module):
         if self.training:
             # Do distillation.
             assert codebook_indices is not None
+            assert torch.all(
+                codebook_indices[torch.where(codebook_indices < 0)] == -100
+            )
+            assert torch.max(codebook_indices) < 256
             assert hasattr(self, "codebook_loss_net")
 
             # Output rate of hubert is 50 frames per second,