add(shentao): add a recent work on segment-level dense RLHF

[Shentao-YANG]

Actually, I only change the following

### 2025
- [Segmenting Text and Learning Their Rewards for Improved RLHF in Language Model](https://arxiv.org/abs/2501.02790)
  - Yueqin Yin, Shentao Yang, Yujia Xie, Ziyi Yang, Yuting Sun, Hany Awadalla, Weizhu Chen, and Mingyuan Zhou
  - Keyword: Segment-level Reward Model, Dense Reward RLHF Framework, Improved PPO training for LLMs
  - Code: [Official](https://github.com/yinyueqin/DenseRewardRLHF-PPO)

Not sure why it shows so many changes...

[PaParaZz1]

Thanks for your contribution, please modify the name of this pull request as previous format. Then we will merge this PR.

For the weird so many changes problem, maybe you can reopen another PR with a new branch based on the latest main branch. This problem maybe be caused by some git operations like git rebase.

[Shentao-YANG]

Thank you for the reply. I modified the name of this pull request.
My PR was already based on the latest main branch at the time of opening it. In case you cannot merge it due to git issues, maybe could you just paste my added item onto the main branch? (It is just 5 lines so this way could save our time.) Thank you!