Beyond GRPO and On-Policy Distillation: An Empirical Sparse-to-Dense Reward Principle for Language-Model Post-Training
Paper • 2605.12483 • Published • 10
XYX
xuyd16
AI & ML interests
None yet
Recent Activity
submitted a paper 15 days ago
Beyond GRPO and On-Policy Distillation: An Empirical Sparse-to-Dense Reward Principle for Language-Model Post-TrainingOrganizations
None yet
authored a
paper 15 days ago
submitted a
paper to Daily Papers 15 days ago
submitted a
paper to Daily Papers about 1 month ago
submitted a
paper to Daily Papers 3 months ago
authored 4
papers 3 months ago
Overconfident Errors Need Stronger Correction: Asymmetric Confidence Penalties for Reinforcement Learning
Paper • 2602.21420 • Published • 6
On-Policy Self-Distillation for Reasoning Compression
Paper • 2603.05433 • Published • 9
Not all tokens are needed(NAT): token efficient reinforcement learning
Paper • 2603.06619 • Published • 1
PACED: Distillation at the Frontier of Student Competence
Paper • 2603.11178 • Published • 4