HealthML / self-supervised-3d-tasks

Apache License 2.0
187 stars 40 forks source link

Pre-training model where to insert path of saved model #19

Closed lokeycookie closed 2 years ago

lokeycookie commented 2 years ago

Hello, I just come across this repo and wanted to try some pre-training on a Brats dataset.

For pre-training, where do I specify the path to where I want the pre-trained model/weights to be saved in? I can't find this path for saving models parameter in the config files.

aihamtaleb commented 2 years ago

Hi, thanks for trying out the repo. That path is created automatically when you start the training. You can find the line here: https://github.com/HealthML/self-supervised-3d-tasks/blob/9f21099867f3ac10fe232bd60899483079fd7c91/self_supervised_3d_tasks/train.py#L51

You need to pass the correct base_workspace as a parameter to train_model. Then you will be able to set the path prefix.

lokeycookie commented 2 years ago

Hello, thank you! One more question, let's say I have run the pre-training and the finetuning files. After the finetuning, where will the model (eg 3d-unet) be saved at? Do we need to specify the path to where this final model is saved at?