CV-OPD: Contrastive Vision On-policy Distillation
To address the issue, we introduce Contrastive Vision On-policy Distillation (CV-OPD), which compares teacher distributions induced by positive and negative crops while conditioning on the same student prefix and augments optimization with a margin-based contrastive constraint, which is modulated with discrepancy gating across token and sample level.
ProofPaper ↗
Key points
- On-policy distillation (OPD) reveals versatility and convenience during VLM post-training, providing dense supervision for student from teacher.
- Specifically, CV-OPD first introduces a negative teacher unrelated to the answer and designs a Contrastive Divergence Loss (CDL) that constrains the feature distribution of student by contrasting the positive and negative teachers, correcting how the student exploits visual regions.
- The CV-OPD introduces a Discrepancy Aware Gate (DAG) that uses token-level and sample-level discrepancies to select visually relevant tokens for optimization, directing the student's attention toward visual content.
- Extensive experiments across a broad range of fine-grained visual benchmarks demonstrate that CV-OPD substantially alleviates shortcut and consistently improves the visual reasoning capabilities of existing VLMs.
Sources (1)
- [1]CV-OPD: Contrastive Vision On-policy DistillationarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 9, 09:15 AM
To address the issue, we introduce Contrastive Vision On-policy Distillation (CV-OPD), which compares teacher distributions induced by positive and negative crops while conditioning on the same student prefix and augments optimization with a margin-based contrastive constraint, which is modulated with discrepancy gating across token and sample level.
On-policy distillation (OPD) reveals versatility and convenience during VLM post-training, providing dense supervision for student from teacher.
Extractive summary: sentences quoted from the sources.
Before this
- Oct 9, 2026Continually Post-training Small Models via Synthetic Self-distillation
- Oct 8, 2026When Do We Need On-Policy Distillation? Distilling on Offline Student Rollouts Is Often Better
- Oct 8, 2026One Block, Multiple Depths: Recurrent Vision Transformers with Depth-Programmed Experts
- Oct 8, 2026SpatialOPSD: Self-Distilling Spatial Intelligence from Verified Coding Agent Traces
- Oct 8, 2026Distilling Routed 3D Privilege for Spatial Reasoning in Vision-Language Models
- Oct 8, 2026ViSkill: Reinforcing VLM Agents with Evolving Visual-Native Skills