H-OPD: Confidence Aware Heterogeneous Multi-Teacher Multimodal On-policy Distillation
arXiv:2607.02592v1 Announce Type: new Abstract: On-policy distillation (OPD) has recently emerged as an effective post-training paradigm by providing supervision on student-generated trajectories. However, existing OPD methods for multimodal reasoning usually rely on a static teacher routing, assigning each sample to a single teacher based on modality or task type. This ignores that visual grounding and abstract reasoning may dominate different decoding steps, making a single teacher insufficien...
arXiv cs.CV
·Qixiang Yin, Huanjin Yao, Yuchen Cai, Jianghao Chen, Ziyi Wang, Min Yang, Fei Su, Zhicheng Zhao
·
// relacionados
Leia também
Editorial
EdgeBench: 38 mil horas de agentes trabalhando revelam uma lei de escala inesperada
Editorial
Hy3: a Tencent libera seu modelo de 295 bilhões sob Apache 2.0 — e sem fronteiras
Editorial
InternVLA-A1.5: o robô que imagina o próximo segundo antes de mover o braço
Editorial