Bonian: The test_DLPFC.ipynb notebook is the original code. The shuffle_DLPFC.ipynb notebook is the one where I used a grid of shuffle ratios to evaluate the model performance. The training procedure used in both notebooks is the same; they both rely on the code in the models.py and networks.py files located in this directory for model construction and training.

Preprocessed DLPFC data are available at [here]