@misc{indiciaeb43540fa65c6,
  title = {Good SFT Optimizes for SFT, Better SFT Prepares for Reinforcement Learning},
  author = {Dylan Zhang and Yufeng Xu and Haojin Wang and Qingzhi Chen and Hao Peng},
  year = {2026},
  url = {https://arxiv.org/abs/2602.01058},
  note = {Source identifier: 2602.01058}
}