Thank you for providing this fantastic project and the detailed codebase. I am currently trying to reproduce your results on the SemanticKITTI dataset and have a few questions about the training process:
Hardware Setup: Could you also provide information about the hardware setup you used during training? For instance:
- The type and number of GPUs used (e.g., RTX 3090, A100, etc.).
- Whether you utilized CPU resources significantly and the specifications if applicable.
- Approximate training time for a single experiment on your setup.
Your insights would be incredibly helpful for my experiments. Thank you for your time and for sharing your amazing work!