Our paper, Scalable Data Ablation Approximations for Language Models through Modular Training and Merging, has been accepted to EMNLP 2024! -> presented in Miami in November