Andy Tang
Co-first author of FRS paper, working at intersection of generative models and robot policy learning at Stanford IRIS Lab.
“FRS is an inference-time mechanism to unlock that latent knowledge without retraining the base model.”
Source→“We use OpenPi's π0.5-LIBERO... For all others, we use π0.5 fine-tuned by Jain et al.”
Source→“up to 95% absolute task success rate boosts in under a minute of training (Abstract) on 10 trajectories — challenges the assumption that adaptation requires significant data collection infrastructure.”
Source→“Authors: Andy Tang, William Chen, Andrew Wagenmaker, Chelsea Finn, Sergey Levine (Stanford + UC Berkeley). Date: June 2025. arXiv: 2606.13675.”
Source→AI-extracted from podcast / newsletter / paper summaries. May contain errors.