Data space dimension processing

[lxwwang]

Hi, the shape of my training data indicates that [fmri.shape]=[(140,64,64,48,1)]. Based on your previous question, would it be more appropriate to pad the data from 64,64,48 to 64,64,64 when processing? However, I hope to load your weights and model after you complete the pre-training; will changing the model input affect the performance of the pre-trained model? I would greatly appreciate it if you could resolve my doubts.