#artificial intelligence
May 2026
Counteraction-Aware Multi-Teacher On-Policy Distillation for General Capability Recovery with Domain Preservation
Counteraction-Aware Multi-Teacher On-Policy Distillation is proposed, which addresses issues with decoupled alternating training and gap-based sample selection and outperforms all baselines in general capability recovery while maintaining domain-specific behavior.
Tian-Le Chen, Jiao Ou, Zi-Yuan Liu et al.
· arXiv.org · 13 citations
· ⚡1