HopeJW

I think you need to look at the source code for bevfusion-pytorch to understand why.

Nv::Tensor is just a data structure that stores a matrix. You can get its raw data pointer. There is nothing special about it.

The quantize.py is mainly to improve the ResNet50-PTQ performance. It would be best if you disabled the PTQ for the SwinT backbone.

it was found that the OSD physical memory was heavily occupied -> Could you expand on this?

Inaccurate measurement of font backend STB font (font set to 10, return height is 12) -> The measurement function is designed to return the size of the area needed for...

DLA requires some special operations. For example: [Deep-Learning-Accelerator-SW](https://github.com/NVIDIA/Deep-Learning-Accelerator-SW)

Sorry, I can't reproduce this error. Could you provide the full code for this issue?

OK, will build a new binary with libc++.

@NoahLinckeKodiak I'm sorry for not getting back to you sooner. I will update it as soon as possible. Thanks.

This ONNX file can only be run by [libspconv.so](https://github.com/NVIDIA-AI-IOT/Lidar_AI_Solution/tree/master/libraries/3DSparseConvolution).