School of EECS, The University of Queensland, Australia
I am a PhD student at The University of Queensland (UQ), Australia, expected to graduate in June 2027. My research focuses on post-training for diffusion large language models (dLLMs), conducted under the supervision of Dr. Ruihong Qiu and Prof. Zi (Helen) Huang.
My publications have been accepted at venues such as NeurIPS, ICML, and WSDM. I completed my dual Bachelor of Computer Science and Master of Data Science degrees at UQ under the supervision of Prof. Zi (Helen) Huang, graduated on 16 December 2023.
We study block-size domain conflict in multi-domain RL post-training for dLLMs, introduce Block-R1-41K and Block-R1 benchmark, and propose sample-level best-improved training block sizes.
We propose b1, a post-training framework that learns dynamic-size reasoning blocks via Monotonic Entropy Descent with reinforcement learning to improve reasoning coherence in dLLMs.
We propose the ALEX framework for balanced and explainable social media analysis in public health using data augmentation and LLM-based explanation.