Develop a model from scratch which can classify the noised mnist dataset with high accuracy. The dataset is noised and given by instructor.
To make a model to have a good performance, image augmentations (flipping, rotation, etc.) are applied to the training dataset.
Develop a model from scratch which translate from language A to language B. The dataset is given by instructor.
To make a model to have a good performance, masking and padding is applied to the training dataset.
Develop a model from scratch which generates the next word based on the given and previous word generated previously.
To make a model to have a good performance, masking and padding is applied to the training dataset.