Jiaxuan You (@youjiaxuan) / X

Jiaxuan You

213 posts

Jiaxuan You

@youjiaxuan

Assist. Prof. @ UIUC (@siebelschool), directing U Lab on LLM Agent infra and application @Stanford CS PhD ex NVIDIA Senior Scientist

Urbana, IL

cs.stanford.edu/~jiaxuan/

Joined April 2017

Pinned
Jiaxuan You
@youjiaxuan
Apr 10
Really thrilled to be part of this! It’s been mind-blowing to move so fast with the team and see our models grow so rapidly. Nothing is impossible when you're working with such brilliant minds. 🥑
Hongyu Ren
@ren_hongyu
Apr 8
Check out Muse Spark, our first milestone in the quest for personal superintelligence! Scaling this with the team has been a total blast. Give it a spin and let us know what you think! 🥑
15K
Jiaxuan You
@youjiaxuan
Dec 14, 2024
Shame on the MIT Media Lab Professor. Racist activities shouldn't have their places at a top AI conference @NeurIPSConf. Imagine changing "Chinese" to other races or ethic groups, she will probably be fired by @MIT. This is not for Chinese. Who knows who will be the next victim?
Jiao Sun
@sunjiao123sun_
Dec 14, 2024
Mitigating racial bias from LLMs is a lot easier than removing it from humans! Can’t believe this happened at the best AI conference @NeurIPSConf We have ethical reviews for authors, but missed it for invited speakers? 😡
47K
Jiaxuan You
@youjiaxuan
Nov 13, 2025
One ICLR reviewer asked us to add 12 NEW experiments—and gave a REJECT score. These aren’t free tweaks; they’re RL training runs for LLMs. Across my lab's submissions, that are 100+ extra experiments if we fully comply. Many requests aren’t even meaningful (“try RL with 8B model
84K
Jiaxuan You
@youjiaxuan
Oct 28, 2025
Introducing Multi-Agent Evolve 🧠 A new paradigm beyond RLHF and RLVR: More compute → closer to AGI No need for expensive data or handcrafted rewards We show that an LLM can self-evolve — improving itself through co-evolution among roles (Proposer, Solver, Judge) via RL — all
72K
Jiaxuan You
@youjiaxuan
May 11, 2025
🤯NeurIPS 2025 might break records as the most submitted-to academic conference ever. One of our submission IDs is already ~23,000 — final count could hit 30,000. Absolute madness. #NeurIPS2025 #AI
121K
Jiaxuan You
@youjiaxuan
Sep 10, 2024
🚀 Our Paper Copilot is now on Arxiv and featured in @ak's daily papers! As a self-evolving LLM RAG system with thought retrieval, personalized profiles, and thread optimization, Paper Copilot offers real-time, personalized research services. Demo: huggingface.co/spaces/ulab-ai…
28K
Jiaxuan You
@youjiaxuan
Nov 18, 2020
We are excited to release #GraphGym, a platform for designing and evaluating #GraphNeuralNetworks. It provides a modularized pipeline, a system for launching thousands of experiments, and more! Code: github.com/snap-stanford/… Paper: arxiv.org/abs/2011.08843 #NeurIPS2020 Spotlight
github.com
GitHub - snap-stanford/GraphGym: Platform for designing and evaluating Graph Neural Networks (GNN)
Platform for designing and evaluating Graph Neural Networks (GNN) - snap-stanford/GraphGym
Jiaxuan You
@youjiaxuan
Jul 14, 2020
How does the graph structure of a neural network affect its predictive performance? Our ICML 2020 paper "Graph Structure of Neural Networks" arxiv.org/abs/2007.06559 reveals many interesting findings on this topic. with Jure Leskovec, Kaiming He, Saining Xie #ICML2020
Jiaxuan You
@youjiaxuan
Nov 12, 2025
Multi-Agent Evolve is now fully open-source 🚀 With our codebase, you can pick your favorite LLM checkpoint and let it self-evolve, WITHOUT external supervision 💻Code: github.com/ulab-uiuc/Mult… 🤗Model Checkpoints: huggingface.co/collections/ul… Feedback and contributions are
Jiaxuan You
@youjiaxuan
Oct 28, 2025
Introducing Multi-Agent Evolve 🧠 A new paradigm beyond RLHF and RLVR: More compute → closer to AGI No need for expensive data or handcrafted rewards We show that an LLM can self-evolve — improving itself through co-evolution among roles (Proposer, Solver, Judge) via RL — all
GitHub - ulab-uiuc/Multi-agent-evolve
From github.com
32K
Jiaxuan You
@youjiaxuan
Aug 8, 2025
RL excels in math & coding tasks with well-defined rewards—but designing rewards for social intelligence is far more challenging. We present Sotopia-RL: Reward Design for Social Intelligence 📄 Paper: arxiv.org/abs/2508.03905 🌐 Project (Code & Model): sotopia.world/projects/sotop…
Haofei Yu
@haofeiyu44
Aug 8, 2025
🚀 RL works wonders for math & coding... But what about social intelligence? We introduce Sotopia-RL: Reward Design for Social Intelligence 🤝🧠 📘 rl.sotopia.world 📑 arxiv.org/abs/2508.03905 🔗 github.com/sotopia-lab/so…
arxiv.org
Sotopia-RL: Reward Design for Social Intelligence
Social intelligence has become a critical capability for large language models (LLMs), enabling them to engage effectively in real-world social tasks such as collaboration and negotiation....
21K
Jiaxuan You
@youjiaxuan
Nov 13, 2025
As an author, reviewer, and AC, I propose adding a resource tag (e.g., industry/academic, low/high compute) to ICLR/ICML/NeurIPS submissions. In the pursuit of AGI, research resources vary drastically. Some works are minimal academic prototypes; others come from well-equipped
22K
Jiaxuan You
@youjiaxuan
Jul 4, 2025
NeurIPS now warns: if your coauthor misses reviews, your paper will be desk rejected Worst part? No option to reassign reviewers Echoes Qin's 连坐 law—punishing people for others' actions Qin fells in 200 BC As former AC, change a reviewer takes 1 min We need inclusion, not fear
25K
Jiaxuan You
@youjiaxuan
Nov 3, 2025
Thrilled that our paper, "Multi-Agent Evolve", was selected as a top AI paper of the week! 📄 Paper: arxiv.org/abs/2510.23595 💻 Code (coming very soon!): github.com/ulab-uiuc/Mult… We propose a new paradigm for LLM self-improvement in post-training, beyond RLHF and RLVR. An LLM
DAIR.AI
@dair_ai
Nov 2, 2025
Top AI Papers of the Week (Oct 27 - Nov 2): - SmolLM2 - AgentFold - Precision-RL - Multi-Agent Evolve - Agent Data Protocol - Graph-based Agent Planning - Introspective Awareness in LLMs Read on for more:
arxiv.org
Multi-Agent Evolve: LLM Self-Improve through Co-evolution
Reinforcement Learning (RL) has demonstrated significant potential in enhancing the reasoning capabilities of large language models (LLMs). However, the success of RL for LLMs heavily relies on...
23K
Jiaxuan You
@youjiaxuan
May 17, 2024
🚀 Excited to announce "How Far Are We From AGI?", first paper summarizing current and future research directions toward #AGI. We hope it inspires researchers to achieve AGI responsibly as a #community. 📄 Paper: arxiv.org/abs/2405.10313 💻 GitHub (PR🤗): github.com/ulab-uiuc/AGI-…
20K