1% of Tokens Can Be Enough: On Gradient Estimation in On-Policy Distillation
Paper • 2609.24432 • Published • 12
winccc
winccc
AI & ML interests
None yet
Recent Activity
upvoted a paper 1 day ago
1% of Tokens Can Be Enough: On Gradient Estimation in On-Policy Distillation upvoted a paper about 1 year ago
Learning to Align, Aligning to Learn: A Unified Approach for
Self-Optimized AlignmentOrganizations
None yet
upvoted a paper 1 day ago