About segmentation config

[MSH970515]

Hello, thank you for great project.

I think I am missing some information about the project.

In segmentation config repository, I can't get the reason for various types of network.

For example, why did you split configs into vssm and vssm1?

Also, in vssm1 dir, what is the difference between ade20k-512x512_tiny and tiny1 excluding depth and ssm_ratio?

Finally, which one should I leverage to get the optimal results?

Thank you.

Best,

[MzeroMiko]

From the contents of those configs, you can easily find out what model is used. In general, vssm stands for vanilla-vmamba, vssm1 stands for vmambav2_*_224.yaml, tiny1 corresponding to classification/configs/vmambav2v_tiny_224.yaml. If you want the fastest, you can choose tiny1. If you want the best performance, I recommend you to build your own config based on the classification checkpoints we released.

[MSH970515]

Thanks for swift reply! This helped me a lot