The Teacher Is a Direction, Not a Destination: Extrapolating RL-Induced Representation Residuals in On-Policy Distillation
Paper • 2609.36484 • Published • 208
E. Kaya
ashishsin
·
AI & ML interests
Data, models, and a bit of chaos.
Recent Activity
upvoted an article 20 days ago
autotrust/JEV-27B: fast, calibrated decisions and full reasoning from one open model upvoted a paper about 2 months ago
Agentic Game Development as a Verifiable Trajectory Data Engine for Scaling World ModelsOrganizations
None yet
upvoted a paper 14 days ago
upvoted an article 20 days ago
Article
autotrust/JEV-27B: fast, calibrated decisions and full reasoning from one open model
autotrust
• • 27