BabyVLM：受婴儿学习启发的 VLM 数据高效预训练
BabyVLM: Data-Efficient Pretraining of VLMs Inspired by Infant Learning
AI总结：
提出 BabyVLM 框架，通过构建领域内评估基准和儿童导向的合成训练数据集，证明在精选数据上训练的紧凑 VLMs 能有效泛化，实现数据高效学习。
摘要：
人类婴儿能从极少输入中迅速发展视觉推理能力，这表明受发育启发的预训练能显著提升视觉语言模型（VLMs）的效率。尽管近期研究利用了像 SAYCam 这样的婴儿启发数据集，但现有评估基准仍存在错配——它们要么过于简单、范围狭窄，要么专为大规模预训练模型定制。此外，仅在婴儿数据上训练忽略了婴儿自然学习所需的更广泛、多样的输入。为解决这些局限，我们提出 BabyVLM，这是一个新颖框架，包含全面的领域内评估基准，以及通过对现有数据集进行儿童导向变换而创建的合成训练数据集。我们证明，与仅在 SAYCam 或同等规模的通用数据上训练的模型相比，使用我们合成数据集训练的 VLMs 在 BabyVLM 任务上取得了更优的性能。因此，BabyVLM 提供了稳健且符合发育阶段的评估工具，并展示了在精心筛选数据上训练的紧凑模型如何有效泛化，为数据高效的视觉语言学习范式开辟了路径。
英文摘要：
Human infants rapidly develop visual reasoning skills from minimal input, suggesting that developmentally inspired pretraining could significantly enhance the efficiency of vision-language models (VLMs). Although recent efforts have leveraged infant-inspired datasets like SAYCam, existing evaluation benchmarks remain misaligned--they are either too simplistic, narrowly scoped, or tailored for large-scale pretrained models. Additionally, training exclusively on infant data overlooks the broader, diverse input from which infants naturally learn. To address these limitations, we propose BabyVLM, a novel framework comprising comprehensive in-domain evaluation benchmarks and a synthetic training dataset created via child-directed transformations of existing datasets. We demonstrate that VLMs trained with our synthetic dataset achieve superior performance on BabyVLM tasks compared to models trained solely on SAYCam or general-purpose data of the SAYCam size. BabyVLM thus provides a robust, developmentally aligned evaluation tool and illustrates how compact models trained on carefully curated data can generalize effectively, opening pathways toward data-efficient vision-language learning paradigms.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
