Filling in the Mechanisms: How do LMs Learn Filler-Gap Dependencies under Developmental Constraints?
填补机制:语言模型如何在发育约束下学习填充-缺口依赖?
机构 * University of Maryland(马里兰大学)
AI总结 研究通过DAS分析语言模型在不同数据量下的填充-缺口依赖表示,发现有限数据下存在共享但敏感的机制,但语言模型仍需更多数据才能达到人类水平,凸显语言特异性偏见的重要性。
Comments To be published in the 64th Annual Meeting of the Association for Computational Linguistics