ml-for-gp/jaxgptoolbox

Geometry processing utilities compatible with jax for autodifferentiation.

jaxgptoolbox's Issues

lip-net in torch

Dear authors,

Thanks for releasing the code! I am wondering if it is possible to also provide the code in pytorch? I want to compare the results with DeepSDF. Thanks a lot for your help and time.

Best,
Xianghui

LipMLP: Xavier vs. He weight initialization used

Hi, congrats on your LipMLP work!
According to the paper appendix A, the weight matrix should be initialized with the He init if ReLU is used.
Regarding the code (https://github.com/ml-for-gp/jaxgptoolbox/blob/main/demos/lipschitz_mlp/model.py#L33)
I was wondering whether the initialization of the weight matrix in the code is correct since ReLU is used: std = sqrt(2 / fan_in) instead of std = gain / sqrt(fan_in).

Request for Requirements Files for Simplified Installation

To facilitate a more efficient installation process, could you please provide a 'requirements.txt' file for pip or a 'requirements.yml' file for conda, if available?

Your assistance is highly appreciated.

Best regards,

Arbab

3D Example?

Hi, first, I love the project.

I was wondering if you might have working code for the PointNet chair example for LipMLP, somewhere, that users can easily try?