Log inSign up
user avatar

Junlin Yang

@junlin45300
1st-year PhD @Tsinghua_Uni advised by @stingning and Bowen Zhou Prev: Research intern @taoyds @haopeng_uiuc
yangjl2003.github.io
Joined April 2024
1,252
Following
565
Followers
RepliesMedia
  • Pinned
    user avatar
    Junlin Yang
    @junlin45300
    Jul 31
    🚀 OpenRSI is a new open research series from @FrontisAI for concrete, testable progress toward recursive self-improvement (RSI). As its first project—and also my first work as first author—I’m proud to present OpenMLE: an open full-stack AI4AI system for autoresearch, where
    Image
  • user avatar
    Junlin Yang
    @junlin45300
    Jul 7
    To me, MLS-Bench is very much in the “AI trains AI” direction: asking whether LMs can help improve AI systems in a setting that feels close to how AI research is actually done today, and testing that question carefully. Check out @Lyubh22 's thread:
    user avatar
    Bohan Lyu
    @Lyubh22
    Jul 6
    Everyone is talking about self-evolving AI, or recursive self-improvement. The methods that built modern ML are the ones that keep working across settings and scales, yet no benchmark directly tests AI systems for that ability. Today, after months of cross-platform validation
    Image
  • user avatar
    Junlin Yang
    @junlin45300
    Jun 25
    Agent world modeling has long been a dream many of us believed in, but making it work was a massive challenge. This work is a systematic step toward world models as simulators and planning tools for agents. Seeing it finally come together is truly exciting. Congrats to the team🙌
    user avatar
    Qwen
    @Alibaba_Qwen
    Jun 24
    📣📣 Meet Qwen-AgentWorld — a native language world model that simulates 7 agent environments (MCP, Search, Terminal, SWE, Web, OS, Android) within a single model. Environment modeling is the training objective from day one, not a post-hoc adaptation. 🤔 LLMs are trained to be
    Image
  • user avatar
    Junlin Yang
    @junlin45300
    Jun 18
    Auto Research is moving fast, but one capability is still under-tested: can an AI scientist update causal beliefs by doing experiments? CausaLab asks agents to observe, intervene, and recover the mechanism—not just predict the answer. 92% task accuracy ≠ causal understanding.
    user avatar
    Dylan Zhang
    @dylan_works_
    Jun 6
    A real scientist doesn't look up how the world works — they intervene, observe, and revise until a theory holds for a case they've never seen. CausaLab drops an LLM agent into a lab where memorized facts are useless ("Quantum Crystals on Planet X") and asks for the same.
  • user avatar
    Junlin Yang
    @junlin45300
    Apr 16
    OPD has been blowing up recently, yet many open questions still remain. What does it really take to make OPD work in practice? This work goes deep into the details — highly recommended and well worth a read!🔥
    user avatar
    Bingxiang He
    @HBX_hbx
    Apr 15
    1/n ✨ Introducing our new work: Rethinking On-Policy Distillation of Large Language Models OPD is now a core technique in LLM post-training (Qwen3, MiMo, GLM-5...). But here's the uncomfortable truth: it often doesn't work. We systematically study the phenomenology, mechanism,
    Image