Questions about the ground-truth poses

[Hui-design]

Thanks for your excellent work.
I observe that your work achieves 99.7% RR in SemanticKitti, where and I have two questions.

1. why you use self-added rotation and translation instead of the ground-truth poses from the Semantic-Kitti. I think registration under the ground-truth is harder and more challenging, which introduces the problem of partial-to-partial registration.
2. why not compare with GeoTransformer, the SOTA in KittiOdometry.