外观
论文arXiv 2610.1886915 天前
质量感知 VLA 预训练中什么重要：一个基准与实证研究
原文链接：What Matters in Quality-Aware VLA Pretraining: A Benchmark and Empirical Study（arXiv 2610.18869 · 发布日期：2026-10-13）
What Matters in Quality-Aware VLA Pretraining: A Benchmark and Empirical Study
研究VLA预训练中如何利用示范质量，比较加权、过滤与条件化，并在AXIS-Bench和真实实验中验证。
核心要点
- 关注目标任务微调后性能，探讨预训练中应如何使用示范质量。
- 在不同奖励定义和标签粒度下，比较加权、过滤、条件化策略。
- 三种VLA中，approach-aware episode条件化较普通行为克隆提升平均成功率。
- AXIS-Bench基于众包仿真示范，支持重算质量标签和可重放轨迹。
验证范围与局限：验证范围包括5000条自然次优人类示范、24个目标任务、三种VLA和真实世界实验。
英文摘要 (Abstract)
How should demonstration quality be used during vision-language-action (VLA) pretraining to improve performance after target-task fine-tuning? We compare weighting, filtering, and conditioning across reward definitions and label granularities using 5,000 naturally suboptimal human demonstrations and 24 target tasks. Within each VLA, we match demonstrations and training budgets. Additional comparisons examine pretraining data size and the roles of quality tags during fine-tuning and inference. Across three VLAs, we find that approach-aware episode conditioning improves mean success over plain behavioral cloning; conditioning and filtering favor episode and chunk labels, respectively, on average, with model-dependent reversals; and constant-tag fine-tuning improves mean success after conditioned pretraining with either tagged or untagged inference. We introduce AXIS-Bench to support these controlled comparisons. Built on crowdsourced simulation demonstrations, it provides recorded object and contact states for recomputing quality labels on shared trajectories. Replayable trajectories support workspace-aligned rendering and user-defined real-world target tasks. In our real-world experiment, simulation pretraining raises success from 9.6% to 31.7%. Project page: https://axis-bench.github.io/.
作者：Mingxuan Yan, Litian Gong, Dihong Huang, Zhixu Li, Tianyu Zhang, Wenqian Zhang, Hefeifei Jiang, Peihao Li, Jian Zhang, Mengfei Zhao
来源
本页事实对应的公开页面，以原文为准。