vector-quantize-pytorch 1.17.2__tar.gz → 1.17.4__tar.gz

{vector_quantize_pytorch-1.17.2 → vector_quantize_pytorch-1.17.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: vector-quantize-pytorch
-Version: 1.17.2
+Version: 1.17.4
 Summary: Vector Quantization - Pytorch
 Project-URL: Homepage, https://pypi.org/project/vector-quantize-pytorch/
 Project-URL: Repository, https://github.com/lucidrains/vector-quantizer-pytorch

{vector_quantize_pytorch-1.17.2 → vector_quantize_pytorch-1.17.4}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "vector-quantize-pytorch"
-version = "1.17.2"
+version = "1.17.4"
 description = "Vector Quantization - Pytorch"
 authors = [
     { name = "Phil Wang", email = "lucidrains@gmail.com" }

{vector_quantize_pytorch-1.17.2 → vector_quantize_pytorch-1.17.4}/vector_quantize_pytorch/finite_scalar_quantization.py

@@ -159,7 +159,6 @@ class FSQ(Module):
 
         return codes
 
-    @autocast('cuda', enabled = False)
     def forward(self, z):
         """
         einstein notation

{vector_quantize_pytorch-1.17.2 → vector_quantize_pytorch-1.17.4}/vector_quantize_pytorch/residual_vq.py

@@ -149,8 +149,11 @@ class ResidualVQ(Module):
         self.quantize_dropout_multiple_of = quantize_dropout_multiple_of  # encodec paper proposes structured dropout, believe this was set to 4
 
         # setting up the MLPs for implicit neural codebooks
+        
+        self.mlps = None
 
-        self.mlps = ModuleList([MLP(dim = codebook_dim, l2norm_output = first(self.layers).use_cosine_sim, **mlp_kwargs) for _ in range(num_quantizers - 1)])
+        if implicit_neural_codebook:
+            self.mlps = ModuleList([MLP(dim = codebook_dim, l2norm_output = first(self.layers).use_cosine_sim, **mlp_kwargs) for _ in range(num_quantizers - 1)])
 
         # sharing codebook logic