vsmae-small | ico4 | 3 frames | 50% masking

Pretrained weights for a vsmae-tiny model pre-trained on t-fMRI data from the HCP dataset with the following parameters: 75% masking ratio, tubelet masking, 3 fMRI frames.