arxiv:2609.25048
🔄 In a Training Loop
Ming Xu (徐明)
shibing624
AI & ML interests
ML, nlp
Recent Activity
authored a paper about 16 hours ago
Teaching a Moving Student: Rethinking the Curriculum of On-Policy Distillation authored a paper about 16 hours ago
When Self-Evolution Backfires: Pre-Commit Gating against Skill Contamination in LLM Agents authored a paper about 16 hours ago
Distillation as Probability Transport: Routed On-Policy DistillationOrganizations
None yet