Biography
Fangru is a DPhil student at University of Oxford co-supervised by Janet Pierrehumbert and Daniel Altshuler. She's also a co-organizer of OxNLP reading group. Her research focuses on using linguistically informed methods to assess and improve state-of-the-art Large Language Models’ performance in real world tasks such as planning and using LLMs as autonomous agents. She is especially interested in the application of neuro-symbolic methods on general artificial intelligence. Her work is supported by Clarendon and Jason Hu studentships. Industrially, she previously served as software engineer summer intern at Microsoft. For more information, please visit her page.
Awards and Prizes
- Clarendon Studentship
- Jason Hu Studentship
- Cohere For AI Research Grant
Research Interests
-
Linguistically informed NLP
-
Neuro-symbolic methods
-
LLMs as autonomous agents
-
Large Multimodal Models
-
Interpretability
Current Projects
- Graph-enhanced Large Language Models in Asynchronous Plan Reasoning, (under review) https://arxiv.org/abs/2402.02805
- Probing Large Language Models for Scalar Adjective Lexical Semantics and Scalar Diversity Pragmatics (under review)
