Scholar
Wen-Tse Chen
Google Scholar ID: VSUDQ0oAAAAJ
Carnegie Mellon University
Citations & Impact
Citations
95
H-index
4
i10-index
2
Publications
9
Co-authors
6
list available
Publications
Resume (English only)
Academic Achievements
- - Proposed Verlog: A multi-turn RL framework for LLM agents, built for long-horizon LLM-agentic tasks with highly variable episode lengths.
- - Presented a sample-efficient method for online fine-tuning LLM agents using in-context learning to convert sparse feedback into dense signals at NeurIPS 2025 (Oral presentation).
- - Introduced DGPO: An on-policy framework for discovering multiple diverse optimal strategies for the same task in a single training process at AAAI 2024.
Research Experience
- Looking for a part-time/summer internship.
Education
- PhD student at Carnegie Mellon University RI, advised by Prof. Jeff Schneider; Undergraduate in Automation at Tsinghua University, worked with Prof. Jun Zhu.
Background
- Research interests focus on LLM agents, deep reinforcement learning, and their applications in decision-making and robotics.
Miscellany
- Page design inspired by Dr. Jon Barron's webpage.
