Can training neural language models on a curriculum with developmentally plausible data improve alignment with human reading behavior?

Can training neural language models on a curriculum with developmentally plausible data improve alignment with human reading behavior? Aryaman Chobey author Oliver Smith author Anzi Wang author Grusha Prasad author 2023-12 text Proceedings of the BabyLM Challenge at the 27th Conference on Computational Natural Language Learning Alex Warstadt editor Aaron Mueller editor Leshem Choshen editor Ethan Wilcox editor Chengxu Zhuang editor Juan Ciro editor Rafael Mosquera editor Bhargavi Paranjabe editor Adina Williams editor Tal Linzen editor Ryan Cotterell editor Association for Computational Linguistics Singapore conference publication chobey-etal-2023-training 10.18653/v1/2023.conll-babylm.9 https://aclanthology.org/2023.conll-babylm.9/ 2023-12 98 111