On-policy distillation (OPD) consolidates expert capabilities into a single student by matching teacher and student final-layer distributions on student-gene…
机构:港理工
来源:arXiv 2610.28048 | AI4Papers 论文推荐平台