This repository contains code and scripts for training and validating TGVNs to replicate https://arxiv.org/abs/2501.03021. Below is an overview of the files and their purposes. As the repository was built on fastMRI, the requirements can be found on /~https://github.com/facebookresearch/fastMRI

• fastmri_split: Contains the data split for the fastMRI knee dataset. Note that the csv files contain absolute paths, so the user should modify them depending on the dataset location.
• m4raw_split: Contains the data split for the M4Raw dataset. Note that the csv files contain absolute paths, so the user should modify them depending on the dataset location.

• Set_I.sbatch: SLURM script for running the first set of experiments using TGVN.
• Set_II.sbatch: SLURM script for running the second set of experiments using TGVN.
• Set_III.sbatch: SLURM script for running the third set of experiments using TGVN.

• custom_losses.py: Implements the custom loss function for training models.
• data.py: Contains data loading and preprocessing logic for fastMRI knee and M4Raw datasets.
• distributed.py: Handles distributed training setup and utilities.
• main_fastmri.py: Main script for training and evaluating models on the fastMRI dataset.
• main_m4.py: Main script for training and evaluating models on the M4Raw dataset.
• models.py: Defines TGVN architecture used in the project.

1. Update the SLURM batch scripts (Set_I.sbatch, Set_II.sbatch, Set_III.sbatch) with the appropriate parameters for your environment.
2. Submit jobs using:

   sbatch Set_I.sbatch
   sbatch Set_II.sbatch
   sbatch Set_III.sbatch

For any questions or issues, feel free to reach out or open an issue in this repository.