Preserving Foundational Capabilities in Flow-Matching VLAs through Conservative SFT

TL;DR

ConSFT is a novel fine-tuning method that preserves pre-trained capabilities in flow-matching VLAs by adaptively scaling learning signals to prevent catastrophic forgetting, without additional data or architecture changes.

Contribution

We introduce ConSFT, a confidence-based optimization approach that maintains pre-trained skills during fine-tuning without requiring prior data or extra network components.

Findings

ConSFT outperforms vanilla SFT in capability retention by over 20%.

ConSFT matches the effectiveness of experience replay without using prior data.

Real-world robotic tests show ConSFT prevents spatial overfitting and preserves physical skills.

Abstract

Unconstrained fine-tuning of flow-matching Vision-Language-Action (VLA) models drives dense parameter overwrites, degrading pre-trained capabilities. We present Conservative Supervised Fine-Tuning (ConSFT), an optimization objective that adapts to target distributions while mitigating catastrophic forgetting, requiring zero prior data or architectural overhead. By dynamically scaling learning signals based on model confidence, ConSFT suppresses excessive gradients from low-confidence samples to prevent disproportionate parameter updates, thereby bounding the intrinsic parameter disruption risk. Inspired by reinforcement learning's trust-region clipping, this formulation establishes a progressive learning dynamic to secure target convergence and prior capability retention, maintaining sparse parameter updates without relying on the parallel reference networks required by explicit regularization. We evaluate ConSFT on the LIBERO and RoboTwin benchmarks across state-of-the-art flow-matching VLAs ($\pi_0$, $\pi_{0.5}$, and GR00T-N1.6-3B). The method outperforms vanilla SFT in capability retention by an average absolute margin of over 20\%, matching the efficacy of data-heavy Experience Replay in a prior-data-free regime. Real-world robotic deployments confirm that ConSFT precludes spatial overfitting during downstream adaptation, preserving pre-trained physical skills while acquiring sequential target tasks.