BA-DPO: Bias-Adjusted Direct Preference Optimization for Language Model AlignmentResearchSep 30, 2026
RS-OPSD: Reliable Privileged On-Policy-Self-Distillation for Ultra-High-Resolution Remote Sensing VQAResearchSep 30, 2026
Towards Understanding On-Policy Distillation through the Lens of Test-Time ScalingResearchSep 30, 2026