On-Policy Distillation (OPD) has emerged as a powerful paradigm for transferring reasoning capabilities from frontier models to smaller models, while extendi…
机构:阿里
来源:arXiv 2610.13238 | AI4Papers 论文推荐平台