// PERSON
Ye Li
ROLE RESEARCHERAT TSINGHUA UNIVERSITYMENTIONS 1LAST SEEN JUNE 2, 2026
// BIO
Ye Li is a PhD student at Tsinghua University, where his research focuses on model efficiency, reinforcement learning, and embodied AI. He is known for work on accelerating vision-language-action models and diffusion policies, including the SP-VLA framework for VLA model acceleration through joint model scheduling and token pruning. His publications also include contributions to adaptive ViT inference, KV cache compression, and offline meta reinforcement learning.
// RECENT MENTIONS
// SIGNALS
1 SIGNAL
01
product·arXiv Physical AI·JUNE 2, 2026
“ElegantVLA provides a general acceleration framework for modern VLA pipelines with explicit action-generation modules, without modifying or retraining the base model.”
Source→AI-extracted from podcast / newsletter / paper summaries. May contain errors.