
Senior Researcher at @MSFTResearch AI for Science, working on reinforcement learning, large language models, and AI for Science.
Cambridge, England
Joined April 2022
- Tweets115
- Following714
- Followers154
- Likes902
Pinned Tweet
Excited to share our latest development, QFANG: A Scientific Reasoning Model for Organic Synthesis Procedure Generation. Check it out 👉arxiv.org/abs/2512.13668
#MicrosoftResearch #AIforScience #Chemistry #ReasoningModel #LLM
Guoqing Liu retweeted
Sharing some of my favorite examples. Proud of our Craft team for building Canva Craft really from scratch in just 5-6 months—the closest model to GPT-Image-2 in performance. This is just the beginning. Don’t hold back—throw your toughest prompts at it! 🚀🚀🚀@canva @DesignArena
Canva Craft, the first ever image model by @canva, takes the #4 spot on DesignArena's Image leaderboard. This makes Canva the #2 Image Lab after @OpenAI!
Notably, it wins most often on photorealistic product and object shots, setting up studio-style lighting and detailed material textures on a single hero object. It also ranks #1 on the Portrait and Landscape subcategories.
Congrats to the @canva team on the launch!
Check out our new model.
Now you can work on magnetic materials like what you do with DFT, input any structure, any spin, converge, and get the results!
github.com/kairosmaterial/pr…
RetroChimera is finally out in Nature. Super excited!
nature.com/articles/s41586-0…
Custom-made molecules are advancing medicine, materials, and agriculture, but producing them is slow and expensive. A new Nature paper highlights RetroChimera, a predictive model that helps accelerate chemical synthesis, helping researchers explore a wide range of molecules.: msft.it/6012a5Ogy
The model weights and code are available on github.com/microsoft/retroch… (MIT license)
2
181
Guoqing Liu retweeted
Synthesis is a major bottleneck in drug discovery. AI synthesis planning can help propose better routes and make generative molecular design tractable.
With our retroChimera model we have been pushing the boundaries of these models further - out in @Nature today. 1/2
Nearly half a year of silence. We spent it studying one problem: how far RL can scale.
MiMo-V2.6 is in the middle of its RL run right now. Three things we scaled: compute (~2B tokens per step, 1568 prompts × 16 rollouts, fully async), environments and harnesses (multi-task agentic RL, mixed across multiple harnesses in one run), and grader compute (agentic in-group credit assignment, with test-case and rubric-based rewards). We'll open-source the details piece by piece over the coming weeks.
Streaming the run: mimo.xiaomi.com/rl/
Guoqing Liu retweeted
PPO had a second wave in the LLM era for reasons unanticipated by the original paper
- the importance-ratio objective fixes biases from numeric error, async training, and forward pass noise
- the clipping objective affects entropy through a mechanism that we didn't know about at the time of publication (DAPO, arxiv.org/abs/2509.26114)
Guoqing Liu retweeted
Excited to announce major #MatterSim updates!
👩🔬 Experimentally synthesized high thermal conductor identified by MatterSim
⚡️ 3-5x inference speed-up
💪 MatterSim-MT: a new multi-task foundation model for in silico materials characterization
⬇️ Details below (1/6)
Guoqing Liu retweeted
Do you want to work on ambitious collaborative research at MSR Cambridge, UK? We are looking for a two-year postdoc role in machine intelligence team at MSR Cambridge. Come work with fantastic colleagues on topics spanning foundational ML research. Apply:
apply.careers.microsoft.com/…
Guoqing Liu retweeted
We are significantly expanding to accelerate our ambitious plans for AI-driven materials discovery at @MSFTResearch AI for Science. Looking for a Data Engineer, ML Engineer and Applied Scientist (UK/NL/DE).
⬇️See job postings below ⬇️
Guoqing Liu retweeted
I don't do podcasts very often - in reality this is my first one ever, but if anyone wants to listen to someone talk about RL for an hour, this is it
How GPT-5 thinks, with @OpenAI VP of Research @MillionInt
00:00 - Intro
01:01 - What Reasoning Actually Means in AI
02:32 - Chain of Thought: Models Thinking in Words
05:25 - How Models Decide How Long to Think
07:24 - Evolution from o1 to o3 to GPT-5
11:00 - The Road to OpenAI: Growing up in Poland, Dropping out of School, Trading
20:32 - Working on Robotics and Rubik's Cube Solving
23:02 - A Day in the Life: Talking to Researchers
24:06 - How Research Priorities Are Determined
26:53 - OpenAI's Culture of Transparency
29:32 - Balancing Research with Shipping Fast
31:52 - Using OpenAI's Own Tools Daily
32:43 - Pre-Training Plus RL: The Modern AI Stack
35:10 - Reinforcement Learning 101: Training Dogs
40:17 - The Evolution of Deep Reinforcement Learning
42:09 - When GPT-4 Seemed Underwhelming at First
45:39 - How RLHF Made GPT-4 Actually Useful
48:02 - Unsupervised vs Supervised Learning
49:59 - GRPO and How DeepSeek Accelerated US Research
53:05 - What It Takes to Scale Reinforcement Learning
55:36 - Agentic AI and Long-Horizon Thinking
59:19 - Alignment as an RL Problem
1:01:11 - Winning ICPC World Finals Without Specific Training
1:05:53 - Applying RL Beyond Math and Coding
1:09:15 - The Path from Here to AGI
1:12:23 - Pure RL vs Language Models
Guoqing Liu retweeted
Prof. Chen Ning Yang, a world-renowned physicist, Nobel Laureate in Physics, Academician of the Chinese Academy of Sciences, Professor at Tsinghua University, and Honorary Director of the Institute for Advanced Study at Tsinghua University, passed away in Beijing due to illness at the age of 103. His life stands as a timeless chapter in human history—one that shines not only for China but for the global community of thinkers and innovators. His legacy will live on forever.
Guoqing Liu retweeted
MLFFs 🤝 Polymers — SimPoly works!
Our team at @MSFTResearch AI for Science is proud to present SimPoly (SIM-puh-lee) — a deep learning solution for polymer simulation.
Polymeric materials are foundational to modern life—found in everything from the clothes we wear and the food we consume to high-performance materials in aerospace, electronics, and medicine. Today, we introduce a new way to simulate them.
We built a machine learning force field (MLFF) to predict macroscopic properties across a broad range of polymers—trained only on quantum-chemical data, with no experimental fitting. Specifically, we accurately compute polymer densities via large-scale MD simulations, achieving higher accuracy than classical force fields. We also capture second-order phase transitions, enabling prediction of glass transition temperatures. These two properties are fundamental to processing and application design. Finally, we created a benchmark based on experimental data for 130 polymers plus an accompanying quantum-chemical dataset—laying the foundation for a fully in silico design pipeline for next-generation polymeric materials.
The incredible team: Jean Helie, @temporaer, Yicheng Chen, Guillem Simeon, @a_kzna, @ErnestoCheco, @erunzzz, Gabriele Tocci, @chc273, @yatao_li, @SherryLixueC, @zunwang_msr, Bichlien H. Nguyen, Jake A. Smith, and Lixin Sun.
📄 Preprint: arxiv.org/abs/2510.13696
⚙️ Data and code release: in progress⏳
#MLFFs #Polymers #AIforScience #DeepLearning #SimPoly #ScientificML #Microsoft #MicrosoftResearch #MicrosoftQuantum
Guoqing Liu retweeted
✨ Internship Opportunity @ Google Research ✨
We are seeking a self-motivated student researcher to join our team at Google Research starting around January 2026. 🚀 In this role, you will contribute to research projects advancing agentic LLMs through tool use and RL, with the goal of enabling breakthrough applications.
We are particularly interested in PhD students with a strong background in these areas. If interested, please send a brief self-introduction and your CV to [email protected]. Looking forward to connecting with talented researchers in this exciting space!
Guoqing Liu retweeted
RetroChimera, now available on Azure AI Foundry, marks a new milestone for predicting synthesis routes to drug-like molecules, opening new possibilities for AI in drug discovery. Learn more: msft.it/6011sutzJ
Guoqing Liu retweeted
Announcing the first workshop on Foundations of Language Model Reasoning (FoRLM) at NeurIPS 2025!
📝Soliciting abstracts that advance foundational understanding of reasoning in language models, from theoretical analyses to rigorous empirical studies.
📆 Deadline: Sept 3, 2025
Guoqing Liu retweeted
Want to join our efforts @MSFTResearch AI for Science to push the frontier of AI for materials? We are the team behind MatterGen & MatterSim and we have 2 job openings! Each can be in Amsterdam, NL, Berlin, DE, or Cambridge, UK. It is a rare opportunity to join a highly talented, collaborative team and build the next frontier model for materials design.
Senior Researcher: jobs.careers.microsoft.com/g…
Senior Research Engineer: jobs.careers.microsoft.com/g…
Guoqing Liu retweeted
I’m building a new team at @GoogleDeepMind to work on Open-Ended Discovery!
We’re looking for strong Research Scientists and Research Engineers to help us push the frontier of autonomously discovering novel artifacts such as new knowledge, capabilities, or algorithms, in an open-ended self-improving loop.
We aim to work on ambitious research projects in a fast-paced manner.
If this sounds appealing to you, apply using the link below by Friday, August 1st EOD:
job-boards.greenhouse.io/dee…
Guoqing Liu retweeted
Scaling up RL is all the rage right now, I had a chat with a friend about it yesterday. I'm fairly certain RL will continue to yield more intermediate gains, but I also don't expect it to be the full story. RL is basically "hey this happened to go well (/poorly), let me slightly increase (/decrease) the probability of every action I took for the future". You get a lot more leverage from verifier functions than explicit supervision, this is great. But first, it looks suspicious asymptotically - once the tasks grow to be minutes/hours of interaction long, you're really going to do all that work just to learn a single scalar outcome at the very end, to directly weight the gradient? Beyond asymptotics and second, this doesn't feel like the human mechanism of improvement for majority of intelligence tasks. There's significantly more bits of supervision we extract per rollout via a review/reflect stage along the lines of "what went well? what didn't go so well? what should I try next time?" etc. and the lessons from this stage feel explicit, like a new string to be added to the system prompt for the future, optionally to be distilled into weights (/intuition) later a bit like sleep. In English, we say something becomes "second nature" via this process, and we're missing learning paradigms like this. The new Memory feature is maybe a primordial version of this in ChatGPT, though it is only used for customization not problem solving. Notice that there is no equivalent of this for e.g. Atari RL because there are no LLMs and no in-context learning in those domains.
Example algorithm: given a task, do a few rollouts, stuff them all into one context window (along with the reward in each case), use a meta-prompt to review/reflect on what went well or not to obtain string "lesson", to be added to system prompt (or more generally modify the current lessons database). Many blanks to fill in, many tweaks possible, not obvious.
Example of lesson: we know LLMs can't super easily see letters due to tokenization and can't super easily count inside the residual stream, hence 'r' in 'strawberry' being famously difficult. Claude system prompt had a "quick fix" patch - a string was added along the lines of "If the user asks you to count letters, first separate them by commas and increment an explicit counter each time and do the task like that". This string is the "lesson", explicitly instructing the model how to complete the counting task, except the question is how this might fall out from agentic practice, instead of it being hard-coded by an engineer, how can this be generalized, and how lessons can be distilled over time to not bloat context windows indefinitely.
TLDR: RL will lead to more gains because when done well, it is a lot more leveraged, bitter-lesson-pilled, and superior to SFT. It doesn't feel like the full story, especially as rollout lengths continue to expand. There are more S curves to find beyond, possibly specific to LLMs and without analogues in game/robotics-like environments, which is exciting.

































