Elamur Research⌕
☰
home / authors / Masahiro Tanaka

Masahiro Tanaka

Masahiro Tanaka has 2 papers in our library, in 2023–2024; often with Chen Liang, Hao Cheng, Liyuan Liu.

2 papers in our library · 2023–2024

OpenAlex · ORCID · h-index 20 · 111 works · 1,179 citations OpenAlex, 2026-10-14

📄 2 works in our library · 🔗 Cited 1,203 times in our library · 1 more in our library without a page yet · ⭐ Paper Stars ranking: #5976

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

🎓Nihon Fukushi UniversityOkuda · Japan2020–2026 · 8 papersEmployment since 2020 · Lecturer, Department of Rehabilitation, Faculty of Health Sciencecurrent🎓Konan UniversityKobe · Japan2006–2023 · 10 papers🎓Gifu UniversityGifu · Japan1998–2025 · 8 papers🎓Kumamoto UniversityKumamoto · Japan2000–2011 · 6 papers🏢Japan TobaccoTokyo · Japan2004–2018 · 5 papers🎓University of TsukubaTsukuba · Japan2000–2015 · 3 papers🎓Kurume UniversityKurume · Japan1999–2011 · 3 papers🎓Fukuoka UniversityFukuoka · Japan2002–2006 · 3 papers🏢MicrosoftRedmond · United States2023–2024 · 2 papers🎓Osaka Dental UniversityHirakata · Japan2013–2020 · 2 papers

From the bylines of their papers (OpenAlex, 2026-10-14), ORCID and Wikidata; institutions on papers are matched by machine. 16 more institutions on single bylines are not listed.

 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).Arts and Humanities · 2 worksComputer Science · 2 worksArtificial Intelligence · 2 worksComputer Vision and Pattern Recognition · 2 worksLanguage and Linguistics · 2 worksHardware and Architecture · 1 work

 Works by year

2020s

20241 paper20231 paper

 Papers

The most cited first (OpenAlex and our library).

  1. DeepSpeed Ulysses: System Optimizations for Enabling Training of Extreme Long Sequence Transformer Models (2023 · 162 citations)
    Standard LLM training parallelism splits work by batch (data parallelism), hidden size (tensor parallelism), or depth (pipeline parallelism).
  2. GRIN: GRadient-INformed MoE (2024 · 7 citations)
    Mixture-of-Experts (MoE) models activate only a few expert subnetworks per token. However, the discrete routing step has no true gradient, so standard training substitutes the gating gradient. Large-scale MoE training also usually relies on expert parallelism, which forces capacity limits and discards overflow…

All works

3 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. DeepSpeed Ulysses: System Optimizations for Enabling Training of Extreme Long Sequence Transformer ModelsarXiv:2309.14509Sam Ade Jacobs, Masahiro Tanaka, Chengming Zhang, et al.2023 · 162 citations
  2. DeepSpeed-FastGen: High-throughput Text Generation for LLMs via MII and DeepSpeed-Inference ↗arXiv 2024 (Microsoft DeepSpeed) · no summary yetConnor Holmes et al.2024 · 78 citations
  3. GRIN: GRadient-INformed MoEarXiv:2409.12136Liyuan Liu, Young Jin Kim, Shuohang Wang, et al.2024 · 7 citations

 Co-authors here

12 of 14 co-authors with a page, the most papers together first.

 Cited by

Papers of our library cite their works 1,203 times (of the 149,439 papers whose bibliographies we read); 176 citing papers with a page here →.

Updated

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