Post-Training, Finetuning, and Steering:A Field Guide for Curating Training Data

arXiv:2610.19143v1 Announce Type: new Abstract: While Vision-Language-Action models have enabled impressive cross-task transfer in zero-shot settings, \emph{adaptation} methods are often necessary to achieve consistent performance on novel tasks. However, even though a wealth of methods for adaptation have been proposed, it is unclear how to choose the best method based on available trajectories. We close this gap by studying the impact of the adaptation method under a variety of common data scenarios: on- and off-policy data, offline vs online adaptation, and whether the data contains optimal or suboptimal trajectories. Our evaluation spans simulated tasks in the LIBERO suite and real tasks on a Franka platform. In the offline data regime, critic-based steering approaches outperform supervised finetuning when failure data is available. However, offline RL techniques for training critic functions fail to outperform simpler Monte-Carlo-based approaches. Finally, in the online setting, critic-based steering is significantly more sample efficient compared to policy-gradient-based finetuning methods, but has a lower performance ceiling than full finetuning.

Abstract:While Vision-Language-Action models have enabled impressive cross-task transfer in zero-shot settings, \emph{adaptation} methods are often necessary to achieve consistent performance on novel tasks. However, even though a wealth of methods for adaptation have been proposed, it is unclear how to choose the best method based on available trajectories. We close this gap by studying the impact of the adaptation method under a variety of common data scenarios: on- and off-policy data, offline vs online adaptation, and whether the data contains optimal or suboptimal trajectories. Our evaluation spans simulated tasks in the LIBERO suite and real tasks on a Franka platform. In the offline data regime, critic-based steering approaches outperform supervised finetuning when failure data is available. However, offline RL techniques for training critic functions fail to outperform simpler Monte-Carlo-based approaches. Finally, in the online setting, critic-based steering is significantly more sample efficient compared to policy-gradient-based finetuning methods, but has a lower performance ceiling than full finetuning.

Robotics (cs.RO)

arXiv:2610.19143 [cs.RO]

(or arXiv:2610.19143v1 [cs.RO] for this version)

https://doi.org/10.48550/arXiv.2610.19143 Focus to learn more arXiv-issued DOI via DataCite (pending registration)

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.

analiza aiqbench

Post-Training, Finetuning, and Steering:A Field Guide for Curating Training Data

Subiecte conexe