The Teacher Is a Direction, Not a Destination: Extrapolating RL-Induced Representation Residuals in On-Policy Distillation
Paper • 2609.36484 • Published • 168
Sandeep Kumar
sandeepkumarsen
·
AI & ML interests
Reading sci-fi, training models, avoiding bugs
Recent Activity
upvoted a paper about 1 month ago
False Frontiers: Diagnosing and Mitigating Co-Cheating in Self-Evolving Search Agents upvoted a paper 2 months ago
StudentSim: Training LLM-based Student SimulatorsOrganizations
None yet