TA-GRPO-d: Trajectory-Aware GRPO for Optimizing Denoising Trajectories in Diffusion LLMs.

Gyunyeop Kim, Sangwoo Kang

Browse the full ACL paper archive.