Author name: Lawrence Feng, Gaurav R. Ghosal, Jacob Mitchell Springer, Ziqian Zhong, Aditi Raghunathan

arXiv:2605.12705v1 Announce Type: new
Abstract: How can we train models whose post-trained capabilities survive subsequent fine-tuning? Rather than focusing on downstream interventions to mitigate forgetting of upstream capabilities, we study how upst…