@article{selectivetimestepweightingandadvantageba,
  title = {Selective Timestep Weighting and Advantage-Based Replay for Sample-Efficient Diffusion RLHF},
  author = {Eric Zhu and Abhinav Shrivastava and Soumik Mukhopadhyay},
  year = {2026},
  eprint = {2607.07693},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2607.07693},
}