James Zou
@james_y_zou
Followers
15K
Following
2K
Media
255
Statuses
1K
@Stanford professor. Chan-Zuckerberg investigator. Sloan Fellow. Overton Prize. @togethercompute. AI for science + health.
Palo Alto, CA
Joined August 2016
What is it like for scientists to work with AI co-scientists? . Excellent @Nature feature discussing researchers' experience interacting with our #VirtualLab and other tools
2
11
86
RT @nazneenrajani: Thanks for sharing our paper, @emollick 🙏. Today's reasoning models fail to effectively allocate test time compute. Our….
0
1
0
RT @emollick: If you want to destroy the ability of DeepSeek to answer a math question properly, just end the question with this quote: "In….
0
215
0
Looking forward to presenting on AI for liver diseases on July 10! Webinar registration info below👇.
Join us next Thursday at Noon EDT for part ☝️ of our three-part webinar series, Intro to #AI in #BioMedicine. Register today for one, two, or all three sessions!
1
3
8
RT @_hanlin_zhang_: [1/n] Discussions about LM reasoning and post-training have gained momentum. We identify several missing pieces:. ✏️Pos….
0
13
0
Great job led by @ShengLiu_ and Tianlang Chen w/ wonderful collaborators!.🌐 See @ShengLiu_'s nice thread for more information:
🧵 1/.🚀 Excited to share our latest work: Fractional Reasoning. We introduce a new way to continuously control the depth of reasoning and reflection in LLM for scaling test time compute, not just switch between “on” and “off” prompts. 💻 Website: #AI
0
1
1
Introducing Fractional Reasoning: a mechanistic method to quantitatively control how much thinking a LLM performs. tldr: we identify latent reasoning knobs in transformer embedding ➡️ better inference compute approach that mitigates under/over-thinking
1
4
11
Introducing Fractional Reasoning: a mechanistic method to quantitatively control how much thinking a LLM performs. tldr: we identify latent reasoning knobs in transformer embedding ➡️ better inference compute approach that mitigates under/over-thinking
3
30
164
RT @ShengLiu_: 🧵 1/.🚀 Excited to share our latest work: Fractional Reasoning. We introduce a new way to continuously control the depth of r….
0
14
0
RT @lx2015: A symposium focusing on AI foundation models, generative models, and agentic AI will be held on Stanford campus before ASTRO an….
0
10
0
RT @togethercompute: Our Open Data Scientist agent is now ranked on the DABStep data analysis leaderboard!. We released everything - so you….
0
5
0
RT @anshulkundaje: @sara_mostafavi (@genentech ) & I (@Stanford) r excited to announce co-advised postdoc positions for candidates with dee….
0
45
0
Minions published in #icml2025! Collaboration protocol between on-device LLM and cloud LLM leads to 80% cost savings🔥 . Updated paper
How can we use small LLMs to shift more AI workloads onto our laptops and phones?. In our paper and open-source code, we pair on-device LLMs (@ollama) with frontier LLMs in the cloud (@openai, @together), to solve token-intensive workloads on your 💻 at 17.5% of the cloud cost
3
6
45
RT @NehzUx: LLMs are getting more powerful, but they still struggle with super long documents. A common trick is "Divide and Conquer" - cho….
0
5
0
Excited to introduce #CollabLLM -- a method to train LLMs to collaborate better w/ humans! Selected as #icml2025 oral (top 1%)🏅. New multi-turn training objective + user simulator👇.
Even the smartest LLMs can fail at basic multiturn communication. Ask for grocery help → without asking where you live 🤦♀️.Ask to write articles → assumes your preferences 🤷🏻♀️. ⭐️CollabLLM (top 1%; oral @icmlconf) transforms LLMs from passive responders into active collaborators.
6
8
50
RT @ShirleyYXWu: Even the smartest LLMs can fail at basic multiturn communication. Ask for grocery help → without asking where you live….
0
44
0
RT @ZainHasan6: Check out our fully open source data science agent! . It performs complex tasks like hyper-parameter optimization, runs exp….
0
3
0
Excited to introduce Open Data Scientist:.✅outperforms Gemini data science agent.✅solves real Kaggle tasks.✅fully open source, easy to adapt.✅sandbox for safe exec.Step-by-step tutorial on building our agent Great job @federicobianchy @ShangZhu18.
1/.We built an open-source AI agent that can reason like a data scientist. It loads data, writes Python code, retrains when models fail, and solves real Kaggle + DABStep tasks. Here’s how we did it (and how you can too): 👇
1
9
82
RT @togethercompute: 1/.We built an open-source AI agent that can reason like a data scientist. It loads data, writes Python code, retrain….
0
30
0