PACT: Perception-Action Causal Transformer for Autoregressive Robotics Pre-Training · HackerPapers discussions

PACT: Perception-Action Causal Transformer for Autoregressive Robotics Pre-Training — mentioned by madratman

Fri, 28 Oct 2022 04:31:26 GMT

Generative self-supervised model for robotics data. Highlights: - Transformer model for autoregressive prediction of states and actions over time to implicitly encode dynamics and behaviors for a particular robot - The representation learned can be fine-tuned to distinct tasks (navigation, mapping, localization) with minimal data - The base transformer model serves as a generative robotics model, similarly to GPT-3, and can be prompted to result in different robot behaviors Paper: https://arxiv.org/abs/2209.11133; Video: https://youtu.be/mNQvQu_atuw; Code and data…

Paper and discussion history. A mention is not an endorsement.

PACT: Perception-Action Causal Transformer for Autoregressive Robotics Pre-Training — mentioned by CheapBreakfast9

Thu, 27 Oct 2022 22:44:41 GMT

Generative self-supervised model for robotics data. Highlights: - Transformer model for autoregressive prediction of states and actions over time to implicitly encode dynamics and behaviors for a particular robot - The representation learned can be fine-tuned to distinct tasks (navigation, mapping, localization) with minimal data - The base transformer model serves as a generative robotics model, similarly to GPT-3, and can be prompted to result in different robot behaviors Paper: https://arxiv.org/abs/2209.11133 Video: https://youtu.be/mNQvQu_atuw Code and data…

Paper and discussion history. A mention is not an endorsement.