Question about the structure of network

[LEEJUHUI6635]

Hello. First of all, thank you for your nice implementation.

What I want to ask is the question why did you implement the network differently from torch-ngp repository. When I understood the torch-ngp code, the input of the rgb network was 15-dim of the feature vector from the sigma network, but in your code, it seems to bring the 16-dim combined with the sigma(1-dim) together. May I know the reason why you implemented it like this?

Thank you for your implementation, again.