Are these two concepts the same idea? or does it make sense to implement codebook with the nn.Embedding class (pytorch)? I know nn.Embedding is used for a trainable lookup t
