Kelvin (Keqiang) Yan
@KeqiangY

RS at Bytedance Seed, Seattle. LLMs for Scientific Discovery. Ex @MSFTResearch @Princeton @PKU1898, etc. Opinions are my own.

sites.google.com/view/keqian…
Joined September 2020
32 Photos and videos
Come work with Yuanqi!
Yuanqi Du
@YuanqiD
Oct 14
We are hiring research interns in the broad area of machine learning and statistics @MSRNE @MSFTResearch! Come work with us! Personally I have very broad interests and I’d love to learn about your favorite problems!
10
984
Interesting RL env analysis
Vals AI
@ValsAI
Oct 7
Xiaomi open-sourced the RL environments behind MiMo v2.6. In two-thirds of the coding tasks, the answer is still in the task's Git history, and MiMo finds it.
180
Will the 25T tokens be released? ☺️
Will Held@WilliamBarrHeld
Sep 23
There’s an enormous amount of open training data on Hugging Face. What does it take to make it work together 🤗? For Marin’s 535B run, we built on 25T tokens from 152 datasets with licenses permitting training. Here’s the work between downloading those and training a model 🧵
Flow diagram of the Marin data pipeline. Starting with 25.25 trillion tokens, deduplication removes 2.13 trillion and benchmark decontamination removes another 13.66 billion, leaving 23.11 trillion. Documents are organized into 40 topics and five quality bands, yielding 200 sampling groups. Results from 872 small-model experiments feed a regression used to choose sampling weights.

ALT Flow diagram of the Marin data pipeline. Starting with 25.25 trillion tokens, deduplication removes 2.13 trillion and benchmark decontamination removes another 13.66 billion, leaving 23.11 trillion. Documents are organized into 40 topics and five quality bands, yielding 200 sampling groups. Results from 872 small-model experiments feed a regression used to choose sampling weights.
3
337
Kelvin (Keqiang) Yan retweeted
Jiaqi Zhang
@JiaqiZhangVic
Sep 9
I’m excited to share that I’ll be joining @Columbia in the CS Department as an Assistant Professor in January 2027, in New York City. My lab will work on advancing causality, AI, experimental design, and biomedical applications. I’ll be taking on PhD students and recruiting postdocs. If you’re interested in working with me, consider applying to the CS PhD program this fall or reach out at jz4255@columbia.edu.
55
88
966
79,928
Will cancer be cured in the near future? 🥰
Weijie Su
@weijie444
Sep 8
Last July, when LLMs won gold medals at the IMO, I felt that someday AI might be able to solve some of the hardest problems in math. Now, an internal OpenAI model (still improving!) has resolved a Millennium Prize Problem. It feels surreal that that day came so, so soon!
1
8
3,219
Kelvin (Keqiang) Yan retweeted
Dr. Kamal Choudhary@dr_k_choudhary
Jul 22
One year as an assistant professor at JHU this month and delighted to share #AtomGPTLab's work has been awarded through both the NSF PCL Test Bed and DOE Genesis programs! Huge thanks to the agencies, colleagues, and our incredible team. More ahead on AtomGPT.org 🚀
2
6
347
Excited to share that I have joined ByteDance Seed AI for Science at Seattle, pushing the boundaries of AI-driven scientific discovery to accelerate real-world breakthroughs. Onward and upward! 😊
2
27
2,794
Kelvin (Keqiang) Yan retweeted
JHU Computer Science@JHUCompSci
Mar 24
Double seminar next Thursday with @Princeton’s @KeqiangY and @UW’s @breda_joe! 🤩 Find more details here: cs.jhu.edu/events/
Computer Science Seminar Series. AI for Materials Discovery: Graphs, Language Models, and Agents. April 2, 2026, 228 Malone Hall. Refreshments available 10:30 a.m. Seminar begins 10:45 a.m. Keqiang Yan, Princeton University.

ALT Computer Science Seminar Series. AI for Materials Discovery: Graphs, Language Models, and Agents. April 2, 2026, 228 Malone Hall. Refreshments available 10:30 a.m. Seminar begins 10:45 a.m. Keqiang Yan, Princeton University.

Computer Science Seminar Series: From Personal to Public Health with Society-Scale Computing Platforms. April 2, 2026, 228 Malone Hall. Refreshments available 12:15 p.m. Seminar begins 12:30 p.m. Joe Breda, University of Washington.

ALT Computer Science Seminar Series: From Personal to Public Health with Society-Scale Computing Platforms. April 2, 2026, 228 Malone Hall. Refreshments available 12:15 p.m. Seminar begins 12:30 p.m. Joe Breda, University of Washington.
1
4
708
I will be at #NeurIPS2025 next week from Wednesday, and I look forward to meeting with old and new friends! Just let me know if you want to talk about AI for Science, AI for Materials Discovery, LLMs for Science, and Agentic systems~😆
1
1
10
1,325
Also open to fun activities if any!
1
669
Received the hard copy of our book on AI for Science from my dear PhD supervisor @ShuiwangJi ✨ Check it out on Foundations and Trends in Machine Learning if interested!
1
18
2,047
I am pleased to share that I will be joining CS @Princeton University as a Postdoctoral researcher. Look forward to working on foundational problems at the intersection of AI, LLMs, and science.
12
4
212
15,770
Kelvin (Keqiang) Yan retweeted
Shuiwang Ji@ShuiwangJi
23 Jul 2025
Our 500+ page AI4Science paper is finally published: Artificial Intelligence for Science in Quantum, Atomistic, and Continuum Systems. Foundations and Trends® in Machine Learning, Vol. 18, No. 4, 385–912, 2025 nowpublishers.com/article/De…
6
23
95
17,326
Kelvin (Keqiang) Yan retweeted
Hongliang Xin@XinGroup_VT
22 Jun 2025
All AIMEDx webinar recordings are available here: piped.video/playlist?list=PL… Thanks to all our speakers and attendees for an incredible season of ideas and inspiration. We are off for a break, wishing everyone a restful summer! #AI4Science #AIMEDx

AIMEDx Webinar Series

Bridging disciplines to ignite learning at the frontier of AI and scientific research.

youtube.com
3
7
1,617
A milestone to be celebrated 🎉 2^10=1,024 citations ✨
13
2
211
15,452
Sincerely thank all my collaborators, and especially my Ph.D. advisor Dr. @ShuiwangJi to make this happen.
3
483
Excited to share our latest preprint on LLM Agents, beyond tool use, for Materials Discovery. 🎉 ✨Unleash the discovery power of LLM agents by human intuitions ✨Achieve SOTA results without training ✨Towards greater autonomy with human in the loop
2
5
21
1,988
Paper is released at arxiv.org/abs/2506.05616

Toward Greater Autonomy in Materials Discovery Agents: Unifying...

We aim at designing language agents with greater autonomy for crystal materials discovery. While most of existing studies restrict the agents to perform specific tasks within predefined workflows,...

arxiv.org
2
492
Kelvin (Keqiang) Yan retweeted
Shuiwang Ji@ShuiwangJi
9 Jun 2025
New paper alert: Toward Greater Autonomy in Materials Discovery Agents: Unifying Planning, Physics, and Scientists. arxiv.org/abs/2506.05616
1
6
33
1,872
Officially Dr. Yan 🥳🎉
8
1
117
8,995
Load more