Dill Lab at USC

The Data, Interpretability, Language and Learning, (DILL) lab, led by Swabha Swayamdipta, explores questions in the intersection of natural language processing and machine learning. Our research considers language (and other) data, models that train on this data and their impact on society. These are some questions we are currently exploring:

1. How can we rigorously evaluate the generative capabilities of language models?
2. Do language models have the ability to generate text with specific properties and what does that reveal about them?
3. What roles do different kinds of data play in the successes or failures of language models?
4. How can we use language models to understand our society better?