The Teacher Is a Direction, Not a Destination: Extrapolating RL-Induced Representation Residuals in On-Policy Distillation
Paper • 2609.36484 • Published • 561
Daniel Torres
dtorres87
·
AI & ML interests
Rock climber. Also writes code.
Recent Activity
liked a model 10 days ago
autotrust/GLM-5.3-Flash-GGUF-DGX-Spark upvoted a paper about 1 month ago
SMELT: Scaling Laws for Compute-Matched MoE Looped TransformersOrganizations
None yet