Elamur Research⌕
☰
home / authors / Ruihua Song

Ruihua Song

Ruihua Song has 4 papers in our library, in 2025, most of them on Language models & pretraining, Reasoning, Vision & language; often with Wayne Xin Zhao, Kun Zhou, Chengxi Li.

4 papers in our library · 2025

OpenAlex · ORCID · Wikidata · h-index 35 · 151 works · 4,505 citations OpenAlex, 2026-10-14

📄 4 works in our library · 🔗 Cited 118 times in our library · 2 more in our library without a page yet · ⭐ Paper Stars ranking: #7355

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

🎓Renmin University of ChinaBeijing · China2019–2026 · 67 paperscurrent🏢Microsoft Research AsiaBeijing · China2003–2020 · 43 papersEmployment since 2003 · Lead Researcher🏢MicrosoftRedmond · United States2009–2020 · 16 papers🎓Shanghai Jiao Tong UniversityShanghai · China2006–2011 · 5 papersEducation 2006–2011 · PhD, Computer Science🎓Tsinghua UniversityBeijing · China1996–2003Education 1996–2000 · BS., Computer Science and TechnologyEducation 2000–2003 · MS., Computer Science and Technology

From the bylines of their papers (OpenAlex, 2026-10-14), ORCID and Wikidata; institutions on papers are matched by machine.

 Fields, topics and concepts

Of its works in our library, with the number of works on each.

📚 Shelves 6The field of science this work belongs to (OpenAlex fields).Computer Science · 4 worksArtificial Intelligence · 4 worksArts and Humanities · 3 worksLanguage and Linguistics · 3 worksComputer Vision and Pattern Recognition · 2 worksComputational Theory and Mathematics · 1 work

🧭 Topics 4Research subjects and languages the work is about.Language models & pretraining · 3 worksReasoning · 2 worksEfficient models & compression · 1 workVision & language · 1 work

 Works by year

2020s

20254 papers

 Papers

The most cited first (OpenAlex and our library).

  1. Towards Effective and Efficient Continual Pre-training of Large Language Models (2025 · 29 citations)
    Continual pre-training (CPT) is a common way to adapt an LLM to new languages or domains. Its main risk is catastrophic forgetting, where the model gains new skills but loses old ones. The authors give a fully open CPT recipe for Llama-3 (8B) that targets two weaknesses: Chinese language ability and multidisciplinary…
  2. Think Silently, Think Fast: Dynamic Latent Compression of LLM Reasoning Chains (2025 · 19 citations)
    Chain-of-Thought prompting improves LLM accuracy on math problems, but it produces long token sequences that are slow and costly to generate.
  3. What Makes for Good Visual Instructions? Synthesizing Complex Visual Reasoning Instructions for Visual Instruction Tuning (2025 · 13 citations)
    The paper asks which properties of visual instruction data actually help Multi-modal Large Language Models (MLLMs) generalize, since earlier collections had never been compared under the same backbone and training setup.
  4. ReForm: Reflective Autoformalization with Prospective Bounded Sequence Optimization (2025 · 8 citations)
    Autoformalization turns natural-language math problems into Lean statements that a machine can check. Current LLMs usually produce code that compiles, but the statements often change the problem's meaning. The authors argue this happens because models translate in one shot and never review their own output. ReForm…

All works

6 works of theirs in our library, the most cited first; 4 with a page here, the others link their original until we summarise them.

  1. User Behavior Simulation with Large Language Model based Agents ↗arXiv:2306.02552 · no summary yetLei Wang et al.2023 · 64 citations
  2. Intelligent Agents with LLM-based Process Automation ↗ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD) · no summary yetYanchu Guan et al.2024 · 60 citations
  3. Towards Effective and Efficient Continual Pre-training of Large Language ModelsAnnual Meeting of the Association for Computational Linguistics (ACL)Jie Chen, Zhipeng Chen, Jiapeng Wang, et al.2025 · 29 citations
  4. Think Silently, Think Fast: Dynamic Latent Compression of LLM Reasoning ChainsConference on Neural Information Processing Systems (NeurIPS)Wenhui Tan, Jiaze Li, Jianzhong Ju, et al.2025 · 19 citations
  5. What Makes for Good Visual Instructions? Synthesizing Complex Visual Reasoning Instructions for Visual Instruction TuningInternational Conference on Computational Linguistics (COLING)Yifan Du, Hangyu Guo, Kun Zhou, et al.2025 · 13 citations
  6. ReForm: Reflective Autoformalization with Prospective Bounded Sequence OptimizationarXiv:2510.24592Guoxin Chen, Jing Wu, Xinjie Chen, et al.2025 · 8 citations

 Co-authors here

12 of 21 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 118 times (of the 149,439 papers whose bibliographies we read); 28 citing papers with a page here →.

Updated

↑ back to top

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