GRPO

Apple Paper Defines Theoretical Limits for Language Model Alignment
Apple researchers have published a NeurIPS paper defining the maximum reward improvement language models can achieve under a fixed KL-divergence budget. The work also…

Snorkel AI: Partial Credit Can Undermine Agent Training, Binary Gating Fixes It
Snorkel AI researchers argue that partial credit, a common way to train AI agents on hard tasks, can push models to satisfy easy checks while abandoning the route to…