Preserving Foundational Capabilities in Flow-Matching VLAs through Conservative SFT

• URL: http://arxiv.org/abs/2605.08879v1
• Date: Sat, 09 May 2026 10:59:03 GMT
• Title: Preserving Foundational Capabilities in Flow-Matching VLAs through Conservative SFT
• Abstract summary: Conservative Supervised Fine-Tuning (ConSFT) is an optimization objective that adapts to target distributions while mitigating catastrophic forgetting.<n>We evaluate ConSFT on the LIBERO and RoboTwin benchmarks across state-of-the-art flow-matching VLAs.
• Score: 12.18095306054314
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• 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 ($π_0$, $π_{0.5}$, 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.

Related papers

• Federated Continual Learning as a Distributed Drift-Plus-Penalty Control Problem [12.498691027665332]
  Federated Continual Learning (FCL) is fundamental to real-world distributed learning systems.<n>FCL requires models to adapt to sequential, non-IID data across clients while mitigating catastrophic forgetting and client drift.<n>We propose Federated Queue-regulated Continual Learning (FedQCL), a framework based on Lynov drift-plus-penalty (DPP) optimization.
  arXiv  Detail & Related papers  (2026-08-21T18:21:19Z)
• FlowCTS: On-policy Continuous Trajectory Supervision of Flow Models [65.32587991232592]
  We propose Flow Continuous Trajectory Supervision (FlowCTS)<n>FlowCTS matches subsequent student and reference trajectories from the same student-visited state.<n>Under a multi-reference setup, single-state FlowCTS-OPD outperforms vanilla KL-based OPD with faster convergence.
  arXiv  Detail & Related papers  (2026-07-27T15:03:22Z)
• FedSQ: Optimized Weight Averaging via Fixed Gating [1.2058208023553034]
  Federated learning enables collaborative training across organizations without sharing raw data.<n>We propose FedSQ (Federated Structural-Quantitative learning), a transfer-d neural federated procedure based on a DualCopy.<n>Fixing the gating reduces learning to within-regime affine refinements, which stabilizes aggregation under heterogeneous partitions.
  arXiv  Detail & Related papers  (2026-04-03T11:54:23Z)
• FOZO: Forward-Only Zeroth-Order Prompt Optimization for Test-Time Adaptation [9.28697795097814]
  Test-Time Adaptation is essential for enabling deep learning models to handle real-world data distribution shifts.<n>Backpropagation-based methods are not suitable for low-end deployment devices.<n>We propose Forward-Only Zeroth-Order Optimization (FOZO), a novel and practical backpropagation-free paradigm for TTA.
  arXiv  Detail & Related papers  (2026-03-05T02:12:48Z)
• TS-Memory: Plug-and-Play Memory for Time Series Foundation Models [63.21390142212087]
  Time Series Foundation Models (TSFMs) achieve strong zero-shot forecasting through large-scale pre-training.<n>Existing solutions face a trade-off: Parametric Adaptation can cause catastrophic forgetting, while Non-Parametric Retrieval improves forecasts but incurs high latency due to datastore search.<n>We propose Parametric Memory Distillation and implement it as TS-Memory, a lightweight memory adapter that augments frozen TSFMs.
  arXiv  Detail & Related papers  (2026-02-12T04:16:19Z)
• Trust-Region Adaptive Policy Optimization [82.09255251747818]
  Post-training methods play an important role in improving large language models' (LLMs) complex reasoning abilities.<n>We introduce TRAPO, a framework that interleavesSupervised Fine-Tuning (SFT) and Reinforcement Learning (RL) within each training instance.<n>Experiments on five mathematical reasoning benchmarks show that TRAPO consistently surpasses standard SFT, RL, and SFT-then-RL pipelines.
  arXiv  Detail & Related papers  (2025-12-19T14:37:07Z)
• Steering Vision-Language-Action Models as Anti-Exploration: A Test-Time Scaling Approach [78.4812458793128]
  We propose textbfTACO, a test-time-scaling framework that applies a lightweight pseudo-count estimator as a high-fidelity verifier of action chunks.<n>Our method resembles the classical anti-exploration principle in offline reinforcement learning (RL), and being gradient-free, it incurs significant computational benefits.
  arXiv  Detail & Related papers  (2025-12-02T14:42:54Z)
• Harnessing Optimization Dynamics for Curvature-Informed Model Merging [17.42364575754576]
  In supervised fine-tuning, multiple capability-based SFT checkpoints must be consolidated into a single model.<n>We introduce Optimization Trajectory Aware (OTA) Merging and Fast Fisher Grafting (FFG)<n>OTA+FFG improves merged-model quality over strong weight-space baselines, reduces negative transfer, and remains robust across sparsity levels.
  arXiv  Detail & Related papers  (2025-09-14T08:59:53Z)
• Adaptive Deadline and Batch Layered Synchronized Federated Learning [66.93447103966439]
  Federated learning (FL) enables collaborative model training across distributed edge devices while preserving data privacy, and typically operates in a round-based synchronous manner.<n>We propose ADEL-FL, a novel framework that jointly optimize per-round deadlines and user-specific batch sizes for layer-wise aggregation.
  arXiv  Detail & Related papers  (2025-05-29T19:59:18Z)
• Towards Continual Learning Desiderata via HSIC-Bottleneck Orthogonalization and Equiangular Embedding [55.107555305760954]
  We propose a conceptually simple yet effective method that attributes forgetting to layer-wise parameter overwriting and the resulting decision boundary distortion. Our method achieves competitive accuracy performance, even with absolute superiority of zero exemplar buffer and 1.02x the base model.
  arXiv  Detail & Related papers  (2024-01-17T09:01:29Z)
• Over-the-Air Federated Learning with Privacy Protection via Correlated Additive Perturbations [57.20885629270732]
  We consider privacy aspects of wireless federated learning with Over-the-Air (OtA) transmission of gradient updates from multiple users/agents to an edge server. Traditional perturbation-based methods provide privacy protection while sacrificing the training accuracy. In this work, we aim at minimizing privacy leakage to the adversary and the degradation of model accuracy at the edge server.
  arXiv  Detail & Related papers  (2022-10-05T13:13:35Z)
• Robust Learning via Persistency of Excitation [4.674053902991301]
  We show that network training using gradient descent is equivalent to a dynamical system parameter estimation problem. We provide an efficient technique for estimating the corresponding Lipschitz constant using extreme value theory. Our approach also universally increases the adversarial accuracy by 0.1% to 0.3% points in various state-of-the-art adversarially trained models.
  arXiv  Detail & Related papers  (2021-06-03T18:49:05Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.