Steering Out-of-Distribution Generalization with Concept Ablation Fine-Tuning
机构 * Harvard University(哈佛大学) ; Northeastern University(东北大学) ; Anthropic ; ML Alignment & Theory Scholars (MATS) program(ML对齐与理论学者(MATS)计划)
高校专区
机构 * Harvard University(哈佛大学) ; Northeastern University(东北大学) ; Anthropic ; ML Alignment & Theory Scholars (MATS) program(ML对齐与理论学者(MATS)计划)
机构 * Northeastern University(东北大学)
Comments Project Page: https://distillation.baulab.info/
机构 * Northeastern University(东北大学) ; United States Military Academy(美国军事学院) ; Sapienza University of Rome(罗马萨皮恩扎大学) ; Purdue University(普渡大学) ; University of Texas, Austin(德克萨斯大学奥斯汀分校)