Bannapol Limanond

Thank you for your quick response. So, during training with the cross-entropy loss, the ground truth is the `one-hot vector `including the ans + "#"? For example, in the dataset,...