search

ha-ha-ha-han / UKBiobank_deep_pretrain

Pretrained neural networks for UK Biobank brain MRI images. SFCN, 3D-ResNet etc.
MIT License
129 stars 36 forks source link

fine-tune problem #3

Closed podismine closed 3 years ago

podismine commented 3 years ago

Hi, han. I'm really interested in your method and I have met some questions. When it goes to fine-tune task with other datasets, the loss gets lower, but the mae or mse holds. I used the kl_loss from your code as the loss function

ha-ha-ha-han commented 3 years ago

Hi Sorry I missed this issue before you close it. Any update? Perhaps it is an implementation problem in MAE/MSE (e.g. one could accidentally feed random labels to MAE/MSE computing functions)