search

rixez / Brats21_KAIST_MRI_Lab

Codes and pretrained weights for winning submission of 2021 Brain Tumor Segmentation (BraTS) Challenge
Apache License 2.0
169 stars 22 forks source link

Stop training to test #10

Closed wxr521314 closed 2 years ago

wxr521314 commented 2 years ago

hi When I was training, I found that after 300 epoch, my dice was basically unchanged, and loss also converged. I wanted to stop training for testing, how should I modify it? Except for setting the training epoch to 300.Is there another way? For example, can I use the results in the middle of training to predict? How do I change the code?

rixez commented 2 years ago

Hi, I believe you can just do Ctrl+c to stop the training. The model with the best validation score is saved as model_best.model in the saved folder so you can just pass that to the argument -chk for the nnUNet_predict command like this:

nnUNet_predict -i <input_folder> -o <output_folder1> -t <TASK_ID> -m 3d_fullres -tr nnUNetTrainerV2BraTSRegions_DA4_BN_BD --save_npz -chk model_best

I hope this works for you!