Jianlong Kwan

Results 7 issues of Jianlong Kwan

According to most cv paper(like vim), they commonly set the d_state=16. But I wonder if d_state=64 should be attached with the number of tokens

Hi! Mentioned that your paper has two Subtract operations which confuses me. Can I just consider them as decouple?

question

Could you please share the code related to the visualization of ERF in OA-CNNS?

![Snipaste_2024-05-29_17-23-41](https://github.com/Pointcept/Pointcept/assets/57404649/d2648aa8-bfc2-4e1b-bbdb-1893921e3dee) Hello! I have a dataset like the image. I want to know how to run in ptv3. Thanks a lot!

I want to use nn.Conv1d instead of causal-conv1d. How should I modify the code?

That's an illuminating work! I've noticed MambaVision uses amp which may cause loss NAN in some vision mamba works. How did you sove this?

I have a 4-stage network, and considering that each stage has a different number of tokens, I want to set different sizes for d_state, e.g., [256, 128, 64, 32]. However,...