Event

Physical Intelligence releases FAST robot action tokenizer for efficient VLA training

Jan 16, 2025 · Research Publication · Physical Intelligence · Foundation Models

  • Physical Intelligence published FAST, an efficient robot action tokenizer designed to connect high-frequency robot action chunks to autoregressive transformer training.
  • The official release says FAST is trained on one million real robot action sequences and enables pi0-FAST policies to train substantially faster while retaining dexterous-control capability.
  • It matters because tokenizer and dataset infrastructure are part of the research substrate behind scalable robotic foundation models.

Physical Intelligence develops foundation models and learning algorithms for robots and other physically actuated systems. Its π-series models are designed to generalize across robot embodiments, tasks, and environments using robot data, language instructions, vision-language-action training, reinforcement learning, and multimodal context conditioning.

Segment
Foundation Models
Event type
Research Publication
Geography
San Francisco, California · United States