Yingxuan Yang
Papers
1
Total Citations
5
H-Index
1
About
Yingxuan Yang is a rising researcher at the forefront of advancing large language model (LLM) agents, with a focused expertise in enhancing their autonomous decision-making and generalization capabilities. Their most notable contribution is the development of TRAD (Thought Retrieval and Aligned Decision), a pioneering framework that introduces step-wise thought retrieval to guide LLM agents through complex tasks like web navigation and online shopping. By aligning retrieved decision trajectories with the agent's current context, TRAD significantly improves performance without the need for extensive fine-tuning, addressing a critical bottleneck in LLM agent deployment. This work, published in 2024, has already garnered 5 citations, signaling its early impact on the field. Yang’s research sits at the intersection of retrieval-augmented generation, in-context learning, and agentic AI, offering practical solutions for building more reliable and adaptable autonomous systems. Their work is particularly valuable for students and researchers exploring how to bridge the gap between static LLM knowledge and dynamic, real-world task execution, making Yang a key voice in the next wave of intelligent agent design.
Research Focus
Key Achievements
Top Papers
- 1