Not All Disagreement Is Learnable: Token Teachability in On-Policy Distillation
Paper • 2605.26844 • Published • 22
Zhang
kongzym
AI & ML interests
None yet
Recent Activity
upvoted a paper 1 day ago
Not All Disagreement Is Learnable: Token Teachability in On-Policy Distillation upvoted a paper 12 days ago
OScaR: The Occam's Razor for Extreme KV Cache Quantization in LLMs and Beyond upvoted a paper 20 days ago
Geometry Conflict: Explaining and Controlling Forgetting in LLM Continual Post-TrainingOrganizations
