- Generation of meaningful inference from highlighted cells.
- Used the Google Totto Dataset for training and testing.
- The model used was T5-Base
- Special tokens were added for seperation between the different cells and rows so that the model too could understand the positions of the input text
- Model was trained for 7 epochs on GPU and later on 12 rpochs on TPU
File Information
- File for data preprocessing -> converting table data into model input format
- GPU Training -> training model on GPU
- TPU Training -> Training model on TPU using Pytorch XLA