arxiv:2609.32722
Yuntai Bao
colored-dye
P(doom) 10%
ยท
AI & ML interests
on-policy distillation, reinforcement learning, mechanistic interpretability, training data attribution
Recent Activity
submitted a paper 6 days ago
Scaling Properties of Same-Family On-Policy Distillation authored a paper 6 days ago
SkillAligner: Treating Retrieved Skills as Adaptable Drafts at Execution Time authored a paper 6 days ago
AttriMem: Attribution-Guided Process Feedback for Agent Memory LearningOrganizations
None yet