Jul 2026
Flux-OPD: On-Policy Distillation with Evolving Contexts
Flux-OPD is proposed, an OPD paradigm that uses evolving contexts as in-training supervision to capture task preferences in open-ended domains and outperforms existing OPD paradigms, highlighting the potential to combine teacher supervision with evolving contexts.
Yu-Ran Wang, Zekun Wang, Bohan Zeng et al.
· arXiv.org · 1 citation