BHC1205

能不能增加读取网页信息的功能？

5

这样可以搭一个简易的搜索功能，扩充模型的知识库； 甚至可以将搜到的结果二次输入给模型，从而实现伪联网功能

Could not find the implementation of the CA module in the code

2

Thank you for your great work. I couldn't find the code implementation for aligning the 3D position between two frames according to ego motion. Could you please point that out?

Is the function of the mask token m to map the gray voxels that are not selected as query in the first stage to the purple voxels in the second...

CBGSDataset

1

Hi, thanks for the great work ! May I ask how much performance improvement can be achieved by enabling CBGS enhancement?

1. I found that augmentation of the point cloud was recorded in [cam_info['rot'] cam_info['tran']](https://github.com/Sense-GVT/Fast-BEV/blob/65c4722bb0147cdcf4054ce1e41ba6361ca278f2/mmdet3d/datasets/pipelines/transforms_3d.py#L1449), but I didn't see where it was used 2. Why not new_transform = transform @ aug_transform...

Multiple view overlaying problem.

3

Thank you for your great work. 1. I was wondering if the overlap area here is not specially treated, but simply covered with the information of the latter view. Is...

官方给出的案例中，conversations最多只有2轮对话；如果我的数据集中，对每张图片的问答有近百轮，将这100个QA做成1个conversations能正常训练么？还是需要进行切分，想直到训练过程中最大能接收1张图片的QA轮数是多少？

I'm very grateful for your outstanding work, and I've noticed that the current code only supports single-GPU training. I'm not sure if I am mistaken or if that is indeed...