FASTer: Toward Efficient Autoregressive Vision Language Action Modeling via Neural Action Tokenization

Open in new window