
Percy Liang
Faculty · Stanford Artificial Intelligence Laboratory (SAIL) · Stanford University
At Prof Rod’s reading desk
Questions to explore
- What should be observable when a model is developed openly?
- How can task adaptation change fewer parameters than full fine-tuning?
Research interests
Percy Liang works on language models, open model development and learning theory.
- Language models
- Open model development
- Learning theory
Prof Rod’s reading notes
Start with a paper
ACL · 2021
Prefix-Tuning: Optimizing Continuous Prompts for Generation
Prof Rod’s reading note: Compare an adaptation method with full fine-tuning: which parameters change, what stays frozen and what evaluation supports the comparison?
Bring a question of your own
Turn reading into research
Sketch which parameters prefix-tuning updates. List the baselines you would need before calling it better than fine-tuning.
Prof Rod’s evaluation lesson helps you ask what was sampled and what a comparison supports. This is our suggested preparation, not the researcher’s admission requirement.
Learn to compare model resultsKeep exploring
More at Stanford University
Explore the universityImprove this page
Suggest a correction, share a paper to start with, or point readers to a group page or opening.