I am a 5th year Ph.D. Candidate in the CLIP Lab at University of Maryland, College Park. My main source of funding is through my research assistantship at the Army Research Lab’s Content Understanding Branch. I am co-advised by Rachel Rudinger of UMCP and Claire Bonial of the Army Research Lab. I have also received substantial guidance from Frank Ferraro

Research Interests

Overall, I am interested in understanding and improving LLM common sense using established linguistic concepts. I am currently focused on developing physical common sense in LLMs for robots to understanding natural language instructions during disaster work.

The main questions I am investigating in my current research are:

1. What level of reasoning do Large Language Models have when it comes to object-based functionality?
2. How can we improve smaller Large Language Models for computationally constrained scenarios?
3. How can we develop LLMs that integrate specific technical knowledge with common sense reasoning?

I have expertise in

• developing ontologies for describing object functionality and state of being
• creating annotator pipelines based on expert-in-the-loop feedback
• developing synthetic datasets with LLMs
• Fine-tuning and evaluating LLMs on low-resource scenarios and tasks

Non-research facts about Mollie

1. I am very good at baking.
2. I am a life-long piano student.
3. I love reading science fiction, especially books written by women and people of color (but also Dune)