The Teacher Is a Direction, Not a Destination: Extrapolating RL-Induced Representation Residuals in On-Policy Distillation
Paper • 2609.36484 • Published • 553
马杰
fangfengmaj
·
AI & ML interests
Chess player, Python enthusiast
Recent Activity
liked a Space 6 days ago
muhammadzahid83/vision_language_similarity liked a model 6 days ago
autotrust/JEV-27B-VL updated a dataset 11 days ago
fangfengmaj/ocean-pointcloud-text-curatedOrganizations
None yet
upvoted 5
papers 11 days ago
UniEvo-VL: An On-policy Self-Distillation Training Recipe for Multimodal Model Self-improvement
Paper • 2609.38721 • Published • 274
False Frontiers: Diagnosing and Mitigating Co-Cheating in Self-Evolving Search Agents
Paper • 2609.39102 • Published • 603
MaLiang-Harness: A Programmable Path to Image and Video Generation
Paper • 2609.34309 • Published • 394
AREX-2: Advancing Self-Improving Agents through Long-Horizon Reflective Tasks
Paper • 2609.38288 • Published • 137