Yuchuan Tian
Titles as the sources gave them. In 2 entries, with 0 checked quotations.
In the Base
Every entry they appear in.
InfraRSI: Harnessing Recursive Self-Improvement for Training Infrastructure
According to the abstract, the InfraRSI Team presents an agent harness that turns human commands into verified optimizations of distributed LLM training, with a five-stage workflow that fixes the task contract to prevent reward hacking.
API-TAIL: Detecting Model Substitution in LLM APIs through Temperature-Aligned Likelihood Tails
According to the abstract, Mei, Tian, Yuan and co-authors describe API-TAIL, an audit meant to detect when an LLM API provider quietly routes some requests to a cheaper model while billing for the advertised one. They say it was evaluated on real commercial APIs.