I added a generator (manual seed) to data.random_split to make data splits consistent across seeds. Previously the training and validation splits were different for each seed causing data leakage and irreproducibility. Documentation: https://pytorch.org/docs/stable/data.html#torch.utils.data.random_split
I added a generator (manual seed) to data.random_split to make data splits consistent across seeds. Previously the training and validation splits were different for each seed causing data leakage and irreproducibility. Documentation: https://pytorch.org/docs/stable/data.html#torch.utils.data.random_split