学会学习：面向低资源视觉-语言建模的逐词动态门控
Looking to Learn: Token-wise Dynamic Gating for Low-Resource Vision-Language Modelling
AI总结：
针对低资源视觉-语言建模，提出带逐词动态门控的轻量解码器架构，在BabyLM挑战赛五个基准上取得竞争性表现，并发现可解释的门控模式，兼顾性能与可解释性。
摘要：
在认知合理的数据量上训练视觉-语言模型，需要重新思考模型如何整合多模态信息。在BabyLM挑战赛2025的视觉赛道约束下，我们提出了一种轻量级基于解码器的架构，其包含：（1）用于自适应融合语言和视觉线索的逐词动态门控，（2）特征调制和通道注意力以最大化有限视觉信息的效用，以及（3）用于视觉接地（visual grounding）的辅助对比目标。在五个基准（BLiMP、BLiMP补充集、EWoK、Winoground和VQA）上的评估显示，其性能与多模态基线相当或更优。更值得注意的是，我们的动态门控在无需显式监督的情况下发现了可解释的模式，倾向于对内容词使用视觉线索，对功能词使用语言线索。虽然我们识别出挑战赛约束中的局限性，例如全局图像嵌入造成的信息瓶颈以及数据集划分导致的训练不稳定性，但我们的发现确立了动态门控作为高效多模态学习的强大工具，即使在严重约束下也能提供可解释性和性能。
英文摘要：
Training vision-language models on cognitively-plausible amounts of data requires rethinking how models integrate multimodal information. Within the constraints of the Vision track for the BabyLM Challenge 2025, we propose a lightweight decoder-based architecture with (1) token-wise dynamic gating for adaptive fusion of linguistic and visual cues, (2) feature modulation and channel attention to maximise the utility of limited visual information and (3) auxiliary contrastive objectives for visual grounding. Evaluation on five benchmarks (BLiMP, BLiMP Supplement, EWoK, Winoground and VQA) shows competitive or superior performance to multimodal baselines. More notably, our dynamic gate discovers interpretable patterns without explicit supervision, favouring visual cues for content words and linguistic cues for function words. While we identify limitations in the Challenge constraints, such as the information bottleneck created by global image embeddings and training instability from the dataset split, our findings establish dynamic gating as a powerful tool for efficient multimodal learning, offering both interpretability and performance even under severe constraints.
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。
