aiqbench analysis
FiberTune addresses a specific collapse issue in vision-language-action models by preserving visual structure along action-equivalent states without incurring inference overhead. The method demonstrates consistent performance gains across multiple benchmarks and architectures, including improvements in long-horizon task success rates. This approach offers a practical training-time objective for enhancing the robustness of VLA policies while maintaining computational efficiency.
FiberTune: Preserving Action-Fiber Visual Residuals in Vision-Language-Action Fine-Tuning
arXiv:2606.08653v1 Announce Type: cross Abstract: Action-supervised fine-tuning of vision-language-action (VLA) policies fits demonstrations effectively but constrains only the directions that change predicted actions, leaving visual structure consistent across action-equivalent states free to collapse. We formalize this as residual visual collapse along local action fibers and propose FiberTune, a training-time objective that preserves teacher-structured visual residuals without adding inference-time overhead. FiberTune uses an online action probe to estimate action-predictive feature directions, filters them from intermediate visual-token representations, and aligns the resulting probe-filtered residuals to a frozen visual teacher while regularizing their effective rank. Under identical training conditions, FiberTune improves over task-loss-only fine-tuning in every one of six controlled simulation settings spanning two benchmarks and two architectures (pi_0.5 and OpenVLA-OFT), as well as on physical SO-101 pick-place; representative gains include +10.7 percentage points SR(5) on long-horizon CALVIN ABC-to-D and physical SO-101 task success rising from 72.7% to 78.1%. Residual diagnostics show that these gains coincide with increased probe-filtered residual teacher alignment and effective rank, consistent with the action-fiber motivation.
Abstract:Action-supervised fine-tuning of vision-language-action (VLA) policies fits demonstrations effectively but constrains only the directions that change predicted actions, leaving visual structure consistent across action-equivalent states free to collapse. We formalize this as residual visual collapse along local action fibers and propose FiberTune, a training-time objective that preserves teacher-structured visual residuals without adding inference-time overhead. FiberTune uses an online action probe to estimate action-predictive feature directions, filters them from intermediate visual-token representations, and aligns the resulting probe-filtered residuals to a frozen visual teacher while regularizing their effective rank. Under identical training conditions, FiberTune improves over task-loss-only fine-tuning in every one of six controlled simulation settings spanning two benchmarks and two architectures (pi_0.5 and OpenVLA-OFT), as well as on physical SO-101 pick-place; representative gains include +10.7 percentage points SR(5) on long-horizon CALVIN ABC-to-D and physical SO-101 task success rising from 72.7% to 78.1%. Residual diagnostics show that these gains coincide with increased probe-filtered residual teacher alignment and effective rank, consistent with the action-fiber motivation.
Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Robotics (cs.RO)
arXiv:2606.08653 [cs.CV]
(or arXiv:2606.08653v1 [cs.CV] for this version)
https://doi.org/10.48550/arXiv.2606.08653 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.
Full text under the publisher's open licence · 267 words
Related coverage
FiberTune: Preserving Action-Fiber Visual Residuals in Vision-Language-Action Fine-Tuning
arXiv · Sep 11, 2026
Links #5: 2026/07
LessWrong · Jul 31, 2026
BORA: Bridging Offline Reinforcement Learning and Online Residual Adaptation for Real-World Dexterous VLA Models
arXiv · Jun 9, 2026
$\Delta$ Flow Matching for Robot Foundation Models
arXiv · Oct 14, 2026