Stanley Wei
Scholar

Stanley Wei

Google Scholar ID: ER8nZwQAAAAJ
Princeton University
Homepage↗Google Scholar↗
Citations & Impact
All-time
Citations
58
 
H-index
3
 
i10-index
2
 
Publications
5
 
Co-authors
0
 
Publications
6 items
Provable Benefits of RLVR over SFT for Reasoning Models: Learning to Backtrack Efficiently
2026
Cited
0
When Errors Can Be Beneficial: A Categorization of Imperfect Rewards for Policy Gradient
2026
Cited
0
Improved high-dimensional estimation with Langevin dynamics and stochastic weight averaging
2026
Cited
0
LiveCodeBench Pro: How Do Olympiad Medalists Judge LLMs in Competitive Programming?
2025
Cited
0
What Makes a Reward Model a Good Teacher? An Optimization Perspective
2025
Cited
0
Provable unlearning in topic modeling and downstream tasks
arXiv.org · 2024
Cited
0
Resume
Academic Achievements
  • Paper 'What Makes a Reward Model a Good Teacher? An Optimization Perspective' accepted at NeurIPS 2025 (Spotlight)
  • Paper 'LiveCodeBench Pro: How Do Olympiad Medalists Judge LLMs in Competitive Programming?' accepted at NeurIPS 2025
  • Paper 'Provable unlearning in topic modeling and downstream tasks' accepted at ICLR 2025
  • Paper 'Transformers Provably Learn Sparse Token Selection While Fully-Connected Nets Cannot' accepted at ICML 2024
  • Paper 'On the Synergy Between Label Noise and Learning Rate Annealing in Neural Network Training' presented at NeurIPS 2023 Workshop on Optimization for Machine Learning
  • NSF Graduate Research Fellowship (2025)
  • Dean's Honored Graduate, UT Austin (2023)
  • 26th Place, ICPC World Finals (2022)
  • Top 100, Putnam Competition (2020, hosted unofficially due to COVID-19)