Your Teacher Can't Help You Here: Combating Supervision Fidelity Decay in On-Policy Distillation
Paper • 2605.30833 • Published •
1
Langlin Huang
shrango
AI & ML interests
LLM Reasoning, Machine Translation
Recent Activity
upvoted a paper 3 days ago
Questioning the Questions: Sustaining Self-Evolution in Reasoning Models upvoted a paper 10 days ago
Hermes: Learning Contextual Reasoning Unlocks Test-Time Scaling upvoted a paper 10 days ago
MILO: Automated Harness Discovery via Orchestrated Multi-Agent EvolutionOrganizations
commented a
paper 3 months ago