Elamur Research⌕
☰
home / authors / Hanze Dong

Hanze Dong

Hanze Dong has 2 papers in our library, in 2023–2024, most of them on Language models & pretraining; often with Jipeng Zhang, Rui Pan, Shizhe Diao.

2 papers in our library · 2023–2024

OpenAlex · ORCID · h-index 11 · 58 works · 378 citations OpenAlex, 2026-10-14

📄 2 works in our library · 🔗 Cited 492 times in our library · 4 more in our library without a page yet · ⭐ Paper Stars ranking: #4092

The profile is the library's match of the name; two people with one name are not always told apart.

Elsewhere: Google Scholar · Semantic Scholar · dblp · ACL Anthology

 Affiliations

🎓Hong Kong University of Science and TechnologyHong Kong · Hong Kong2019–2026 · 20 papersEducation 2019–2023current🏢SalesforceSan Francisco · United States2023–2026 · 11 paperscurrent🏢MicrosoftRedmond · United States2025–2026 · 4 paperscurrent🎓University of Illinois Urbana-ChampaignUrbana · United States2023–2025 · 8 papers🎓Fudan UniversityShanghai · China2015–2023 · 4 papersEducation 2015–2019🏢MicrosoftSingapore · SingaporeEmployment

From the bylines of their papers (OpenAlex, 2026-10-14), ORCID and Wikidata; institutions on papers are matched by machine. 3 more institutions on single bylines are not listed.

 Concepts, shelves and topics

Of its works in our library: the number of works on each and its rating among them.

📚 Shelves 6The field of science this work belongs to (OpenAlex fields).Artificial Intelligence · 2 worksArts and Humanities · 2 worksComputer Science · 2 worksLanguage and Linguistics · 2 worksSafety Research · 1 workSocial Sciences · 1 work

🧭 Topics 1Research subjects and languages the work is about.Language models & pretraining · 1 work

 Works by year

2020s

20241 paperMitigating the Alignment Tax of RLHF20231 paperRAFT: Reward rAnked FineTuning for Generative Foundation Model Alignment

 Papers

The most cited first (OpenAlex and our library).

  1. RAFT: Reward rAnked FineTuning for Generative Foundation Model Alignment (2023 · 239 citations)
    Aligning generative foundation models with human preferences usually relies on RLHF with PPO. PPO is unstable and sample-inefficient, and it must keep several large models in memory at once (policy, reference, reward, critic). The authors propose RAFT (Reward rAnked FineTuning), a simpler alternative that repeats…
  2. Mitigating the Alignment Tax of RLHF (2024 · 82 citations)
    RLHF makes LLMs follow human preferences, but it also tends to erode skills learned in pre-training. This loss is called the alignment tax. The authors measured it by aligning OpenLLaMA-3B on the Anthropic helpfulness/harmlessness data with Rejection Sampling Fine-tuning (RSF) and DPO. They then tracked performance…

All works

6 works of theirs in our library, the most cited first; 2 with a page here, the others link their original until we summarise them.

  1. RAFT: Reward rAnked FineTuning for Generative Foundation Model AlignmentarXiv:2304.06767Hanze Dong, Wei Xiong, Deepanshu Goyal, et al.2023 · 239 citations
  2. RLHF Workflow: From Reward Modeling to Online RLHFarXiv:2405.07863 · original ↗ · no summary yetHanze Dong et al.2024 · 106 citations
  3. Iterative Preference Learning from Human Feedback: Bridging Theory and Practice for RLHF under KL-ConstraintarXiv:2312.11456 · original ↗ · no summary yetWei Xiong et al.2023 · 93 citations
  4. Mitigating the Alignment Tax of RLHFConference on Empirical Methods in Natural Language Processing (EMNLP)Yong Lin, Hangyu Lin, Wei Xiong, et al.2024 · 82 citations
  5. MLLM-Protector: Ensuring MLLM's Safety without Hurting PerformanceProceedings of the 2024 Conference on Empirical Methods in Natural Language Processing · original ↗ · no summary yetRenjie Pi et al.2024 · 57 citations
  6. Entropy-Regularized Process Reward ModelarXiv:2412.11006 · original ↗ · no summary yetHanning Zhang et al.2024 · 17 citations

 Co-authors here

9 co-authors with a page, the most papers together first.

Venues

Where their works here were published, the most works first, with the editions of those that have a page.

 Cited by

Papers of our library cite their works 492 times (of the 141,840 papers whose bibliographies we read); 30 citing papers with a page here →.

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/authors/hanze-dong/