Gemini Pretraining @GoogleDeepMind, Computational Linguist
London, England
Joined January 2024
- Tweets217
- Following562
- Followers7.4K
- Likes343
Absolutely fortunate to join the stellar team 💫
I did say we'd be releasing our models periodically...introducing Gemini 4 Carbon, our next frontier model.
Following Argon's release just two weeks ago, Carbon shows the pace at which we’re pushing the frontier, it ranks #1 on the recently updated v5.0 of @ArtificialAnlys Intelligence Index at 59. It also leads across key cybersecurity evaluations scoring 88.2% on SEC-bench Pro for finding vulnerabilities, 54.8% on ExploitGym for exploitation, and 71.0% on CWE-bench for patching.
We’re seeing significant improvements in long-horizon tasks even compared to Argon, and we’ve strengthened frontier safeguards.
As we did with Argon, we’re going to roll it out gradually + responsibly. Some trusted cyber defenders will get access starting this week. We'll be expanding it to more partners through our Fairwind Program, ahead of wider availability soon, so hang tight!
I have started a full time Research Scientist position to work on Gemini Large Scale Pretraining at @GoogleDeepMind! I’m based in London and super excited to catch up with old friends and meet new friends!
Fangru Lin retweeted
‼️ Hiring AI Safety Researchers ‼️
I am moving to Berlin to start a new research group on AI safety and societal impacts @HPI_DE. From September this year, I will be hiring fully funded PhDs, postdocs, and research interns.
My group will study the risks and benefits that emerge as we integrate highly capable AI systems into the social and political processes our societies depend on (e.g. communication, coordination, competition). We will measure these impacts through model evals, computational experiments, human studies, and develop technical mitigations. We will do important research that matters and have fun doing it.
Ideal candidates have a background somewhere between CS/NLP and the social sciences, are excited about working on AI in the current moment, and care deeply about making a positive impact through open research. If that sounds like you, PLEASE GET IN TOUCH using the form in the post below.
I am sad to be leaving @oiioxford and @AISecurityInst but will try my very best to bring the amazing energy of these places and their people to Germany. I am also very grateful to the DFG / Emmy Noether programme for funding my group for the next 6 (!) years.
Please spread the word, and I look forward to hearing from you 🤗
🚨Happening now at Hall A poster 2513! Stop by if you are interested in the paper! #ICML2026
I'm presenting procedure generalization at @icmlconf main and @WiMLworkshop on 8th July! Have you ever wondered how LLMs generalize from code to complex language problem-solving? TLDR: RL curriculum is the key!
As usual, I have a pink poster which is impossible to miss🎀!
My paper thread here! Stop by tomorrow if you are interested in how LLMs can/cannot transfer knowledge from symbolic (e.g., code) to natural language problem solving! I’ll be with a pink poster 🎀!
Our paper on LLM isomorphic generalisation is accepted at #ICML2026! Current LLMs are heavily trained on symbolic data like code, but whether/how they generalize to natural language problem solving? 🧵
I’m looking for RS/MTS/RE positions! DM me if you are interested in my work!
My list (tried and good!):
1. Torriden hydration mask: mild and effective hydration and super cheap!
2. Longtake shampoo: smells great and works perfectly on my hair!
3. Nothing better pumpkin juice: it makes me feel healthy 😆
Fangru Lin retweeted
I won't be at #ICML2026 — but our poster will be.
Surprising finding: the SFT checkpoint that scores best can become the worst after identical RL.
"Good SFT Optimizes for SFT, Better SFT Prepares for Reinforcement Learning"
I'm presenting procedure generalization at @icmlconf main and @WiMLworkshop on 8th July! Have you ever wondered how LLMs generalize from code to complex language problem-solving? TLDR: RL curriculum is the key!
As usual, I have a pink poster which is impossible to miss🎀!
You can see a blogpost about this paper:
fangru-lin.github.io/analogy…
, and the paper link is:
arxiv.org/abs/2602.03542, Twitter thread: xitter.cf/FangruLin99/status/205….
Our paper on LLM isomorphic generalisation is accepted at #ICML2026! Current LLMs are heavily trained on symbolic data like code, but whether/how they generalize to natural language problem solving? 🧵
I’m looking for RS/MTS/RE positions! DM me if you are interested in my work!
I'm open to full-time positions! Happy to talk if you are hiring! My CV: fangru-lin.github.io/cv/
Super interesting observation! I wonder if controlling structural diversity could help? We measure abstract structure representations in an ICML paper this year: xitter.cf/FangruLin99/status/205…, and more relevant works are on the way!
wrote up some random experiments I did playing around w/ absolute zero at the start of the year: ivison.id.au/2026/05/06/self…
a little negative which I attribute mainly to skill issues on my part but potentially interesting to some :)
Our paper on LLM isomorphic generalisation is accepted at #ICML2026! Current LLMs are heavily trained on symbolic data like code, but whether/how they generalize to natural language problem solving? 🧵
I’m looking for RS/MTS/RE positions! DM me if you are interested in my work!
I have experience working at frontier labs and multiple first author publications on LLM reasoning/agents. Happy to chat if you are hiring! My CV link: fangru-lin.github.io/cv/
Paper link: arxiv.org/abs/2602.03542
While humans can often generalize across domains in few shot or even zero shot, language models do not do so. Can LLMs make key scientific breakthroughs given that?
More details in the linked thread. Look forward to presenting and discussing about this idea in Seoul!
