Visualize: Figure 1 & 3 & 5 & 6.
Description: "Questions in our datasets involve reasoning with rules. The inputs to the model are the context (facts + rules) and a question. The output is the T/F answer to the question"