Closed shengyu27 closed 3 months ago
I have reviewed the pre training weights provided by the author and found that the depth is: Small: [2,2,27,2] (downsampling) Tiny: [2,2,9,2] (downsampling) This is not consistent with [2,2,2,2] in the paper?
@shengyu27 Hello, VM-UNet only utilizes partial pre-trained weights, which you can configure in config.py.
I have reviewed the pre training weights provided by the author and found that the depth is: Small: [2,2,27,2] (downsampling) Tiny: [2,2,9,2] (downsampling) This is not consistent with [2,2,2,2] in the paper?