Two new papers Discrete Adjoint Schrödinger Bridge Sampler and Rethinking the Design Space of Reinforcement Learning for Diffusion Models: On the Importance of Likelihood Estimation Beyond Loss Design available on arXiv.