Xin Zhou comments

Results 35 comments of


                                            Xin Zhou

Question about `fc_feature = fc_feature * (scale + 1) + shift` in head.py.

> Hi, > > This is used for time embedding. We directly follow the original Diffusion model implementation: https://github.com/lucidrains/denoising-diffusion-pytorch/blob/main/denoising_diffusion_pytorch/denoising_diffusion_pytorch.py#L179-L180. Thanks for your reply! From my point of view, in DDPM,...

Problems when installing pointnet2_ops

Thank you for your interest in our work and apologies for the delay in response. I have not encountered a similar issue, but you could consider updating the `pip` version.

Problems when installing pointnet2_ops

I am closing this issue. Please feel free to reopen it if necessary.

Excuse me, if this box is not checked, does it mean that the code for splitting and splitting has not been released yet?

Yes.

Excuse me, if this box is not checked, does it mean that the code for splitting and splitting has not been released yet?

We are still working on it.

Excuse me, if this box is not checked, does it mean that the code for splitting and splitting has not been released yet?

I am closing this issue. Please feel free to reopen it if necessary.

The question of pre-training accuracy

Thank you for your interest in our work! The code currently open-sourced from the repository aligns with the paper's current version. Our models are all trained and tested using a...

The question of pre-training accuracy

> May I ask if you can directly test with the weights you provide? The accuracy of the test in pointmamba is only 7%~8%, and the weight files provided by...

The question of pre-training accuracy

> 并且当我训练的时候会nan 同样的数据，在mae中可以正常训练请问可以直接使用代码中的超参数吗，需要针对性调整？ I'm sorry, we never encountered an issue with NaN.

The question of pre-training accuracy

I am closing this issue. Please feel free to reopen it if necessary.