Reinforcement Learning Finetunes Small Subnetworks in Large Language Modelsarxiv.org3 points·s-macke··0 commentsOpen articleSaveView on HN