DriveGPT: Scaling Autoregressive Behavior Models for Driving

Xin Huang, Eric M Wolff, Paul Vernaza, Tung Phan-Minh, Hongge Chen, David S Hayden, Mark Edmonds, Brian Pierce, Xinxin Chen, Pratik Elias Jacob, Xiaobai Chen, Chingiz Tairbekov, Pratik Agarwal, Tianshi Gao, Yuning Chai, Siddhartha Srinivasa
Proceedings of the 42nd International Conference on Machine Learning, PMLR 267:25908-25921, 2025.

Abstract

We present DriveGPT, a scalable behavior model for autonomous driving. We model driving as a sequential decision-making task, and learn a transformer model to predict future agent states as tokens in an autoregressive fashion. We scale up our model parameters and training data by multiple orders of magnitude, enabling us to explore the scaling properties in terms of dataset size, model parameters, and compute. We evaluate DriveGPT across different scales in a planning task, through both quantitative metrics and qualitative examples, including closed-loop driving in complex real-world scenarios. In a separate prediction task, DriveGPT outperforms state-of-the-art baselines and exhibits improved performance by pretraining on a large-scale dataset, further validating the benefits of data scaling.

Cite this Paper

BibTeX
@InProceedings{pmlr-v267-huang25ak, title = {{D}rive{GPT}: Scaling Autoregressive Behavior Models for Driving}, author = {Huang, Xin and Wolff, Eric M and Vernaza, Paul and Phan-Minh, Tung and Chen, Hongge and Hayden, David S and Edmonds, Mark and Pierce, Brian and Chen, Xinxin and Jacob, Pratik Elias and Chen, Xiaobai and Tairbekov, Chingiz and Agarwal, Pratik and Gao, Tianshi and Chai, Yuning and Srinivasa, Siddhartha}, booktitle = {Proceedings of the 42nd International Conference on Machine Learning}, pages = {25908--25921}, year = {2025}, editor = {Singh, Aarti and Fazel, Maryam and Hsu, Daniel and Lacoste-Julien, Simon and Berkenkamp, Felix and Maharaj, Tegan and Wagstaff, Kiri and Zhu, Jerry}, volume = {267}, series = {Proceedings of Machine Learning Research}, month = {13--19 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v267/main/assets/huang25ak/huang25ak.pdf}, url = {https://proceedings.mlr.press/v267/huang25ak.html}, abstract = {We present DriveGPT, a scalable behavior model for autonomous driving. We model driving as a sequential decision-making task, and learn a transformer model to predict future agent states as tokens in an autoregressive fashion. We scale up our model parameters and training data by multiple orders of magnitude, enabling us to explore the scaling properties in terms of dataset size, model parameters, and compute. We evaluate DriveGPT across different scales in a planning task, through both quantitative metrics and qualitative examples, including closed-loop driving in complex real-world scenarios. In a separate prediction task, DriveGPT outperforms state-of-the-art baselines and exhibits improved performance by pretraining on a large-scale dataset, further validating the benefits of data scaling.} }
Endnote
%0 Conference Paper %T DriveGPT: Scaling Autoregressive Behavior Models for Driving %A Xin Huang %A Eric M Wolff %A Paul Vernaza %A Tung Phan-Minh %A Hongge Chen %A David S Hayden %A Mark Edmonds %A Brian Pierce %A Xinxin Chen %A Pratik Elias Jacob %A Xiaobai Chen %A Chingiz Tairbekov %A Pratik Agarwal %A Tianshi Gao %A Yuning Chai %A Siddhartha Srinivasa %B Proceedings of the 42nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2025 %E Aarti Singh %E Maryam Fazel %E Daniel Hsu %E Simon Lacoste-Julien %E Felix Berkenkamp %E Tegan Maharaj %E Kiri Wagstaff %E Jerry Zhu %F pmlr-v267-huang25ak %I PMLR %P 25908--25921 %U https://proceedings.mlr.press/v267/huang25ak.html %V 267 %X We present DriveGPT, a scalable behavior model for autonomous driving. We model driving as a sequential decision-making task, and learn a transformer model to predict future agent states as tokens in an autoregressive fashion. We scale up our model parameters and training data by multiple orders of magnitude, enabling us to explore the scaling properties in terms of dataset size, model parameters, and compute. We evaluate DriveGPT across different scales in a planning task, through both quantitative metrics and qualitative examples, including closed-loop driving in complex real-world scenarios. In a separate prediction task, DriveGPT outperforms state-of-the-art baselines and exhibits improved performance by pretraining on a large-scale dataset, further validating the benefits of data scaling.
APA
Huang, X., Wolff, E.M., Vernaza, P., Phan-Minh, T., Chen, H., Hayden, D.S., Edmonds, M., Pierce, B., Chen, X., Jacob, P.E., Chen, X., Tairbekov, C., Agarwal, P., Gao, T., Chai, Y. & Srinivasa, S.. (2025). DriveGPT: Scaling Autoregressive Behavior Models for Driving. Proceedings of the 42nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 267:25908-25921 Available from https://proceedings.mlr.press/v267/huang25ak.html.

Related Material