Scalable Data Ablation Approximations for Language Models through Modular Training and Merging
Paper • 2410.15661 • Published • 1
Datasets and models for EMNLP paper "Scalable Data Ablation Approximations for Language Models through Modular Training and Merging"