Regarding errors in the process of reproducing the code

[qstongxue]

Hello, I tried to reproduce your code, but I keep encountering errors. I noticed that in the experiment.py file, you use torch.save(value, os.path.join(argv.targetdir, 'attention', str(k), f'{key}.pth')) to save the REST results as .pth files. However, in the analysis.py file, you attempt to load them using .npy format with node_attention = np.concatenate([np.load(os.path.join(argv.targetdir, 'attention', str(k), 'node_attention.npy')) for k in range(argv.k_fold)]). How can I resolve this issue? This mismatch seems like it would cause problems in subsequent plotting.