paper-with-me

홈 › Papers

Counteraction-Aware Multi-Teacher On-Policy Distillation for General Capability Recovery with Domain Preservation

2026-05-26 · Tianlei Chen, Jiao Ou, Ziyuan Liu, Ruiming Tang, Jian Liang, Han Li arxiv

Domain specialization can improve LLM behavior in vertical domains, but often weakens the general capabilities inherited from the original model. Recent Multi-Teacher On-Policy Distillation (MOPD) pipelines recover model capabilities by supervising student-generated trajectories with teacher feedback, but typically assume teacher-aligned prompt coverage, requiring prompts to match the teachers' training distributions. This assumption is difficult to satisfy when the general teacher is an open-source model whose post-training data are unknown. Instead of attempting to reconstruct this hidden distribution, we study general capability recovery with readily available proxy general prompts. We identify two failure modes of vanilla MOPD in this incomplete-coverage situation: recovery-preservation counteraction from mixing conflicting recovery and preservation gradients, and weak-signal flattening from uniformly averaging samples with unequal correction demand. We propose Counteraction-Aware Multi-Teacher On-Policy Distillation (CaMOPD), which addresses these issues with decoupled alternating training and gap-based sample selection. CaMOPD gives general recovery dedicated updates, periodically reviews domain prompts for preservation, and selects samples with larger averaged token-level teacher-student log-probability gaps to concentrate correction signals. Across role-play dialogue and medical reasoning QA scenarios, CaMOPD performs best in general recovery over baselines while maintaining domain-specific behavior. Gradient coherence analyses further support the intended effect of CaMOPD in producing more coherent correction signals.

📄 PDF Abstract BibTeX arXiv:2605.27115

Code (0)

등록된 구현이 없습니다.

Similar Papers 제목 키워드 기반

Reinforcement-aware Knowledge Distillation for LLM Reasoning

2026-02-26 · Zhaoyang Zhang, Shuli Jiang, Yantao Shen, Yuting Zhang 외 arxiv

Reinforcement learning (RL) post-training has recently driven major gains in long chain-of-thought reasoning large language models (LLMs), but the high inference cost of such models motivates distillation into smaller st…

Knowledge DistillationReinforcement Learning

H-OPD: Confidence Aware Heterogeneous Multi-Teacher Multimodal On-policy Distillation

2026-07-01 · Qixiang Yin, Huanjin Yao, Yuchen Cai, Jianghao Chen 외 arxiv

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 …

Multimodal ReasoningVisual Grounding

Multi-Label Knowledge Distillation

2023-08-12 · ICCV 2023 1 · Penghui Yang, Ming-Kun Xie, Chen-Chen Zong, Lei Feng 외

Existing knowledge distillation methods typically work by imparting the knowledge of output logits or intermediate feature maps from the teacher network to the student network, which is very successful in multi-class sin…

Binary ClassificationKnowledge DistillationMulti-Label Learning

Multi-Turn On-Policy Distillation with Prefix Replay

2026-07-06 · Baohao Liao, Hanze Dong, Christof Monz, Xinxing Xu 외 arxiv

We study on-policy distillation (OPD) for agentic tasks, where an LLM agent interacts with an environment over multiple turns and a student imitates a teacher over these multi-turn interaction histories. Fully online OPD…

Mathematical Reasoning

Beyond Imitation: Filtering On-Policy Distillation by Reasoning Progress

2026-08-19 · Chen Yang, Haiyuan Wan, Rengrong Xiong, Yize Chen 외 arxiv

On-policy distillation (OPD) has emerged as an effective framework for post-training language models by pairing student-generated trajectories with dense token-level supervision from a teacher. However, OPD implicitly as…