/ML4NLP

Primary LanguageJupyter Notebook

ML4NLP

Notebook "Generate Perturbed Inputs," creates the proposed input perturbation data by executing the cells. The generated data can be used directly as inputs to the neural QA models we analyze. To do this, you need to manually set up each one of them.

Notebooks "VQA" and "SQUAD" include inside all the necessary information in order to run them, and thus generate attributions. To run "TQA," you need to manually set up the Neural Programmer system.