Double-Z

The linearRegression.fit() is too slow.

1

https://github.com/mit-han-lab/amc/blob/040d83fe9f0288556c8afe363c190e763425a6fe/lib/utils.py#L130

Hi, this is a great work, but I have a question: have you compared your spconv's speed with [tf3d](https://github.com/google-research/google-research/tree/master/tf3d)? Thanks!

I am sure I have loaded pretrained resnet50 weights from torchvision to do pruning. the original acc: `top1: 83.1091, top5: 95.0389, loss: 0.6449` My training scripts: ```shell python amc.py --arch=resnet50...

I'm using grid_map in my project, and I want add pcl::PointXYZI data type in the layer, but grid_map only support for float type. I know this is for other function...

About sematic segmentation data class.

1

I have visualized semantic annotation as below. The black points including class: `Other Static Object`, but I think many of them can be class into `building` or `pole`. 

Question about the post processing.

1

Hi, TextField is a great work, but I'm confused about the post processing: > We apply a simple dilation δ (with 3 × 3 structuring element) to group the representatives...

In the original papper section **2.3. Efficient Implementation**, the hash tabel lookup is most time decreasing module, how can I use this? Have anybody make it?

Will future carla use pybind11 as Python API?

4

Hi, Carla is a great work! Carla uses `boost.python` as python api, and there is another python-c++ api [`pybind11`](https://github.com/pybind/pybind11). As far as I know, pybind11 is more pure and simple,...