Skip to content

Reproduce Baselines

Reproducing baselines used in this study is very similar to the adaptation of the FM as essentially we are just using the FM weights for adaptation.

Randomly initialized

We provide the YAML configuration to train the random init baseline at experiments/baselines/supervised_training/supervised_random_init.yaml

By default, we configure this for Task 1. You can adapt this for Task 2 and Task 3 by searching for Note: comments in the YAML that outline what must be changed.

You can start training by running this in the root code folder,

lighter fit --config_file ./experiments/baselines/supervised_training/supervised_random_init.yaml

Transfer learning

We provide the YAML configuration to train the transfer learning baseline at experiments/baselines/supervised_training/supervised_finetune.yaml

This baseline is only used for Task 2 and Task 3 as we use the random init baseline from Task 1 for the transfer. Follow the Note: comments to switch between Task 2 and Task 3 configurations.

You can start training by running this in the root code folder,

lighter fit --config_file ./experiments/baselines/supervised_training/supervised_finetune.yaml

Med3D / MedicalNet

Original repo: https://github.com/Tencent/MedicalNet

We have provided re-implementations of Med3D to fit into our YAML workflows at experiments/baselines/med3d/finetune.yaml. Again, the Note: comments help adapt for different tasks.

You can start training by running this in the root code folder,

lighter fit --config_file ./experiments/baselines/med3d/finetune.yaml

Models Genesis

Original repo: https://github.com/MrGiovanni/ModelsGenesis

We have provided re-implementations of Models Genesis to fit into our YAML workflows at experiments/baselines/models_genesis/finetune.yaml. Again, the Note: comments help adapt for different tasks.

You can start training by running this in the root code folder,

lighter fit --config_file ./experiments/baselines/models_genesis/finetune.yaml