Scholar
Yuan Yang
Google Scholar ID: Lt4tmL8AAAAJ
Georgia Institute of Technology
Citations & Impact
Citations
744
H-index
11
i10-index
13
Publications
20
Co-authors
15
list available
Publications
Browse publications on Google Scholar (top-right) ↗
Resume (English only)
Academic Achievements
- - Project CW: Casually working on LLM-based cyber waifu/husbando
- - The Language Model OS: Introduced a compressor-retriever architecture for lifelong context management
- - Can LLMs Reason in the Wild: Introduced the 'reasoning in the wild' task to evaluate LLMs' ability to solve unknown type reasoning problems
- - LLM for Natural Language to First-Order Logic Translation: Created a high-quality sentence-level NL-FOL pair dataset and proposed an SFT+RLHF framework to fine-tune LLaMA models
- - Inductive Reasoning in Temporal Data: Developed a reasoning framework that detects inductive patterns in temporal data via neural-logic methodology
- - Auto Data-Labeling Framework for ML Models: Studied the problem for graph reasoning models and proposed a learning-by method
Research Experience
- - Research Scientist at ByteDance
- - Focuses on developing new LLM architectures that manage long-term context and function like an intelligent OS
- - Works on LLM-based data synthesis and enhancing LLMs with better reasoning and tool-using capabilities
- - Previous research involved inductive and deductive logic reasoning on knowledge graphs and frameworks for data labeling and adversarial defense
Education
- - Ph.D. in Machine Learning, 2024, Georgia Institute of Technology, supervised by Prof. Faramarz Fekri and Prof. Le Song
- - MS in Computational Data Science, 2017, Carnegie Mellon University
- - BEng in Software Engineering, 2016, Beihang University
Background
- Yuan Yang is a research scientist at ByteDance. His research interests focus on LLM (large language models), logic reasoning, and interpretable and data-efficient machine learning models.
