# Breaking Lock-In: Preserving Steerability under Low-Data VLA Post-Training

arXiv: https://arxiv.org/abs/2604.23121 · HTML: https://mlsanity.com/paper/2604.23121/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- Placed in VLA architectures and generalization, one of 13 directions in Vision-language-action models.
- Builds on 57 of its 60 references that are in the map.
- First posted on arXiv 25 Apr 2026.

## Placed in

- Vision-language-action models / [VLA architectures and generalization](https://mlsanity.com/vla/vla-architectures-and-generalization/)

## Builds on (most cited first, 30 of 57)

- [On the Opportunities and Risks of Foundation Models](https://mlsanity.com/paper/2108.07258/) (2021)
- [RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic Control](https://mlsanity.com/paper/2307.15818/) (2023)
- [OpenVLA: An Open-Source Vision-Language-Action Model](https://mlsanity.com/paper/2406.09246/) (2024)
- [π0: A Vision-Language-Action Flow Model for General Robot Control](https://mlsanity.com/paper/2410.24164/) (2024)
- [Octo: An Open-Source Generalist Robot Policy](https://mlsanity.com/paper/2405.12213/) (2024)
- [LIBERO: Benchmarking Knowledge Transfer for Lifelong Robot Learning](https://mlsanity.com/paper/2306.03310/) (2023)
- [GR00T N1: An Open Foundation Model for Generalist Humanoid Robots](https://mlsanity.com/paper/2503.14734/) (2025)
- [PaLM 2 Technical Report](https://mlsanity.com/paper/2305.10403/) (2023)
- [Open X-Embodiment: Robotic Learning Datasets and RT-X Models](https://mlsanity.com/paper/2310.08864/) (2023)
- [DROID: A Large-Scale In-The-Wild Robot Manipulation Dataset](https://mlsanity.com/paper/2403.12945/) (2024)
- [Robust fine-tuning of zero-shot models](https://mlsanity.com/paper/2109.01903/) (2021)
- [PaliGemma: A versatile 3B VLM for transfer](https://mlsanity.com/paper/2407.07726/) (2024)
- [FAST: Efficient Action Tokenization for Vision-Language-Action Models](https://mlsanity.com/paper/2501.09747/) (2025)
- [Evaluating Real-World Robot Manipulation Policies in Simulation](https://mlsanity.com/paper/2405.05941/) (2024)
- [Dataless Knowledge Fusion by Merging Weights of Language Models](https://mlsanity.com/paper/2212.09849/) (2022)
- [LIBERO-Plus: In-depth Robustness Analysis of Vision-Language-Action Models](https://mlsanity.com/paper/2510.13626/) (2025)
- [A Survey on Vision-Language-Action Models for Embodied AI](https://mlsanity.com/paper/2405.14093/) (2024)
- [Zero-Shot Robotic Manipulation with Pretrained Image-Editing Diffusion Models](https://mlsanity.com/paper/2310.10639/) (2023)
- [LIBERO-PRO: Towards Robust and Fair Evaluation of Vision-Language-Action Models Beyond Memorization](https://mlsanity.com/paper/2510.03827/) (2025)
- [RL's Razor: Why Online Reinforcement Learning Forgets Less](https://mlsanity.com/paper/2509.04259/) (2025)
- [Steering Your Diffusion Policy with Latent Space Reinforcement Learning](https://mlsanity.com/paper/2506.15799/) (2025)
- [Ctrl-World: A Controllable Generative World Model for Robot Manipulation](https://mlsanity.com/paper/2510.10125/) (2025)
- [SimpleVLA-RL: Scaling VLA Training via Reinforcement Learning](https://mlsanity.com/paper/2509.09674/) (2025)
- [Spatial Forcing: Implicit Spatial Representation Alignment for Vision-language-action Model](https://mlsanity.com/paper/2510.12276/) (2025)
- [Improving Vision-Language-Action Model with Online Reinforcement Learning](https://mlsanity.com/paper/2501.16664/) (2025)
- [RoboCat: A Self-Improving Generalist Agent for Robotic Manipulation](https://mlsanity.com/paper/2306.11706/) (2023)
- [Steering Your Generalists: Improving Robotic Foundation Models via Value Guidance](https://mlsanity.com/paper/2410.13816/) (2024)
- [XSkill: Cross Embodiment Skill Discovery](https://mlsanity.com/paper/2307.09955/) (2023)
- [In-Context Imitation Learning via Next-Token Prediction](https://mlsanity.com/paper/2408.15980/) (2024)
- [Contrastive Region Guidance: Improving Grounding in Vision-Language Models without Training](https://mlsanity.com/paper/2403.02325/) (2024)

## Built on by (most cited first, 6 of 6)

- [What Matters in Orchestrating Robot Policies: A Systematic Study of Hierarchical VLA Agents](https://mlsanity.com/paper/2606.10267/) (2026)
- [Flatness Preserves Instruction Following in Vision-Language-Action Models](https://mlsanity.com/paper/2606.23641/) (2026)
- [Position: Vision-Language-Action Models Cannot Be Verified to Perform Physical Reasoning](https://mlsanity.com/paper/2606.30686/) (2026)
- [Causeway: Restoring Task Accessibility for Instruction Switching in VLA Policies](https://mlsanity.com/paper/2609.30913/) (2026)
- [SAMBAR: Selective Anchoring via Method of Multipliers for Balanced Knowledge Acquisition and Retention in Vision-Language-Action Models](https://mlsanity.com/paper/2609.32108/) (2026)
- [When Listening Becomes Easier: Scrubbing Visual Cues for Shortcut-Free VLAs](https://mlsanity.com/paper/2610.10912/) (2026)

## Abstract (arXiv)

Have you ever post-trained a generalist vision-language-action (VLA) policy on a small demonstration dataset, only to find that it stops responding to new instructions and is limited to behaviors observed during post-training? We identify this phenomenon as lock-in: after low-data, supervised fine-tuning (SFT), the policy becomes overly specialized to the post-training data and fails to generalize to novel instructions, manifesting as concept lock-in (fixation on training objects/attributes) and spatial lock-in (fixation on training spatial targets). Many existing remedies introduce additional supervision signals, such as those derived from foundation models or auxiliary objectives, or rely on augmented datasets to recover generalization. In this paper, we show that the policy's internal pre-trained knowledge is sufficient: DeLock mitigates lock-in by preserving visual grounding during post-training and applying test-time contrastive prompt guidance to steer the policy's denoising dynamics according to novel instructions. Across eight simulation and real-world evaluations, DeLock consistently outperforms strong baselines and matches or exceeds the performance of a state-of-the-art generalist policy post-trained with substantially more curated demonstrations.