ReSPO: Reshaped Sequence Policy Optimization for Gradient Starvation in Off-Policy Learning Paper • 2609.35433 • Published 18 days ago • 15
Post-Training Frontier Text-to-Image Models by Composing Preference and Rubric Rewards Paper • 2610.02967 • Published 14 days ago • 41
Rethinking Automated Voice Similarity by Shifting from EER to Embedding Geometry Paper • 2609.33999 • Published 19 days ago • 6
ReCAST: Reward Credit Assignment across Timesteps for Online Diffusion Reinforcement Paper • 2609.13425 • Published Sep 11 • 10