PhD @ NJU (LAMDA) /
Visiting @ UCSD (Hao AI Lab)
I'm interested in the efficiency and infrastructure of LLMs.
· Speculative Decoding OnlineSPEC · AdaFlash
· JetSpec
· PD Survey
· Diffusion LLMs d3LLM
· RLHF & RSI OnlineRLHF · Counterfactual Distillation

