training problem

[zhh6425]

Hi, thanks for your great jobs.

May I know your GPU devices and training details?

[chrockey]

Hi,

Sorry for the late reply.
For semantic segmentation on S3DIS, you need 4GPUs with 24GB memory.
For shape classification and part segmentation, you can probably train the model with a single GPU with 12GB memory.
We also have added configuration files for training details in the master branch.

Regards,

Chunghyun