This repository provides the released post-training actor checkpoints for
PolicyTrim, a two-stage reinforcement learning framework for improving the
intrinsic policy efficiency of Vision-Language-Action (VLA) models.
Most deployment-efficiency methods reduce the latency of each model forward
pass. PolicyTrim instead reduces how many inference calls and physical actions
are required to finish a task. It targets two policy-level bottlenecks:
unreliable predictions near the tail of an action chunk;
redundant physical execution steps and corrective actions.
PolicyTrim first extends the reliable executable action horizon, then applies a
redundancy-aware step-saving objective with stability regularization. Across
three benchmarks and three VLA model families, the method reports:
3x improvement in action chunk utilization;
51.4% reduction in physical execution steps;
up to 5.83x end-to-end deployment speedup;
no compromise in task success rates.
For the method, training code, configuration files, and evaluation scripts, see
the PolicyTrim GitHub repository.
Checkpoint restoration depends on the matching VLA backend and distributed
training configuration. Follow the setup and evaluation instructions in the
GitHub README, then
point the corresponding PolicyTrim configuration to the downloaded checkpoint.
License
The released materials are provided under the
Apache License 2.0.
Users are also responsible for complying with the licenses and terms of the
corresponding base VLA models, datasets, and simulation environments.
Citation
If you find PolicyTrim useful, please cite:
bibtex
1@inproceedings{policytrim2026,
2 title = {PolicyTrim: Boosting Intrinsic Policy Efficiency of Vision-Language-Action Models},
3 author = {Xianghui Wang and Feng Chen and Wenbo Zhang and Hua Yan and Zixuan Wang and Changsheng Li and Yinjie Lei},
4 booktitle = {European Conference on Computer Vision (ECCV)},
5 year = {2026}
6}
