Closed hummat closed 2 months ago
MLPs in ResidualVQ are always initialized, even if implicit_neural_codebook=False.
implicit_neural_codebook=False
https://github.com/lucidrains/vector-quantize-pytorch/blob/54d29e8fba72443a29928e9f94ee52ca43eeb60a/vector_quantize_pytorch/residual_vq.py#L153
@hummat oops, yes indeed
thank you Matthias!
MLPs in ResidualVQ are always initialized, even if
implicit_neural_codebook=False
.https://github.com/lucidrains/vector-quantize-pytorch/blob/54d29e8fba72443a29928e9f94ee52ca43eeb60a/vector_quantize_pytorch/residual_vq.py#L153