Research project
Agent Confidence Estimation
A research framework for studying confidence and uncertainty in language model agents.
- Role
- Undergraduate researcher · UCSC IRKM Lab
- Year
Work
At UC Santa Cruz’s IRKM Lab, I built Agent Confidence Estimation (ACE), a framework for studying uncertainty in language model agents.
Approach
ACE can interchange models, datasets, and prompts. I implemented confidence baselines using semantic similarity across generated response sets.
Availability
Research results are not public. My résumé documents the work.