Models for all subtaks are saved and to reproduce their results, run notebooks Task_A_Test.ipynb, Task_B_Test.ipynb, Task_A_Test.ipynb respectively.
For training from scratch run Task_A_Training.ipynb, Task_B_Training.ipynb, Task_C_Training.ipynb respectively.
Note: Subtask A results for training from scratch may not be completely reproducible because we had to save and load model in the middle of training due to the hardware limitation.