toyot-li

PAA training error

6

Thanks for your error report and we appreciate it a lot. **Checklist** 1. I have searched related issues but cannot get the expected help. 2. I have read the [FAQ...

@sshaoshuai Hello, I wonder what is the motivation for redefining a `DistributedSampler` [here](https://github.com/open-mmlab/OpenPCDet/blob/master/pcdet/datasets/__init__.py#L27), whose `__iter__` method is pretty similar to the PyTorch built-in DistributedSampler, as shown in the PyTorch source...

@sshaoshuai Thanks for the awesome toolbox. But I'm not quite clear about why [this line](https://github.com/open-mmlab/OpenPCDet/blob/master/tools/train.py#L154) here occurs before the DDP model wrapper, even with the simple comment. Could you explain...

Question about `detach`

7

Hi, Thanks for the nice work. For calculating the IoU target, I think the `detach` should be used on the "predicted items". Specifically, in my view, for these two lines...

How to generate mmdet3d_nuscenes_2f_infos_train/val.pkl

2

@yingfei1016 The tools/generate_sweep_pkl.py corresponds to pkl with 30f, but how to generate 2f counterparts? Thanks.

About the details of pre-trained VoVNet checkpoints

5

Hello @yingfei1016 , I am a little confused about the pre-trained VoVNet checkpoints. [This link provided by DETR3D](https://drive.google.com/drive/folders/1h5bDg7Oh9hKvkFL-dRhu5-ahrEp2lRNN) provides a dd3d_det_final.pth, that is first trained on DDAD15M, then trained on...

@yinchimaoliang Thanks for your great work. The EMA update is formulated as $model_{EMA} = d \times model_{EMA} + (1-d) \times model_{current}$, where $d=0.9999 \times (1- e^{-(\text{iters} \times \text{epochs} / 2000)})$....

evaluation errors

10

When doing the evaluation, the following error accurs: ```bash tensorflow.python.framework.errors_impl.InvalidArgumentError: 2 root error(s) found. (0) INVALID_ARGUMENT: activation input is not finite. : Tensor had NaN values [[{{node 2905231348_876199450/conv_2/CheckNumerics}}]] [[strided_slice_2/_5]] (1)...

error when running av2 dataset

7

@LegendBC Hello, when running MapTR on the AV2 dataset. It goes into https://github.com/hustvl/MapTR/blob/main/projects/mmdet3d_plugin/datasets/av2_map_dataset.py#L627, then the training stops. Could you please provide some suggestions on solving the problem? Thanks.

why there is a re-implementation of gradient checkpointing

3

HI @afterthat97, thanks for your awesome work. I noticed there is a `checkpoint.py` file that seems to re-implement `torch.utils.checkpoint`. What is the motivation for that? Furthermore, could you please provide...