Hello, I would like to ask you what is the matter with light_training, I am doing 3D medical image segmentation, using A40 training, using two SegMamba as my backbone segmentation network, bs=4 burst memory, bs=2 I do not know if I can achieve the optimal results, can there be a lightweight mamba can be used?
Hello, I would like to ask you what is the matter with light_training, I am doing 3D medical image segmentation, using A40 training, using two SegMamba as my backbone segmentation network, bs=4 burst memory, bs=2 I do not know if I can achieve the optimal results, can there be a lightweight mamba can be used?