This model combines the base version of distilroberta + the standalone version of LiLT. It was created with the code available at the original LiLT repository
https://github.com/jpWang/LiLTThe model can be used for fine-tuning in token classification tasks or visual question answering.
