Post-Training Layer Scaling Prevents Forgetting and Enhances Model Mergingarxiv.org1 point·veryluckyxyz··0 commentsOpen articleSaveView on HN