Yuan comments

Results 3 comments of


                                            Yuan

Models trained in v0.6.3 give wrong output in v0.7.0

I got same problem with my model, after comparing 0.6.3 and git version, the difference comes from https://github.com/lukemelas/EfficientNet-PyTorch/blob/3d400a58023086b5c128ecd4b3ea46c129b5988b/efficientnet_pytorch/model.py#L199 and https://github.com/lukemelas/EfficientNet-PyTorch/blob/3d400a58023086b5c128ecd4b3ea46c129b5988b/efficientnet_pytorch/model.py#L204 in version 0.6.3 uses `global_params.image_size` only, and all Conv2d are...

谢谢作者的时间，我运行了代码后出现这样的错误

这个代码是在PyTorch0.4上跑的，在新版本中要在lr_scheduler.step() 之前调用 optimizer.step()。不过似乎这只是一个警告，代码还是一样运行

cuda12 support for spconv v1.1.2

the cuda 12 version has been upload here: https://github.com/NVIDIA-AI-IOT/Lidar_AI_Solution/tree/master/libraries/3DSparseConvolution/libspconv/lib/x86_64 however the results are off: ![cuda-bevfusion_cuda12](https://github.com/NVIDIA-AI-IOT/Lidar_AI_Solution/assets/239020/9f93aeb3-08b3-4cb7-8280-0cb4d0292b4b) Comparing to cuda 11 version: ![cuda11-bevfusion_r50_fp16](https://github.com/NVIDIA-AI-IOT/Lidar_AI_Solution/assets/239020/11092105-4e6b-45b5-b004-1d054a1c663c) @hopef could you fix this? thx!