Lin Yang (@lyang36) / X

Lin Yang

206 posts

Lin Yang

@lyang36

Associate Professor of ECE&CS@UCLA. ML, RL, big data, algorithms, astronomy.

Los Angeles, CA

Joined October 2011

Pinned
Lin Yang
@lyang36
Aug 27, 2025
Our IMO gold medal-winning AI pipeline is now model-agnostic. 🥇 What worked for Gemini 2.5 Pro now gets the same 5/6 score with GPT-5 & Grok4. This confirms the power of our verification-and-refinement pipeline to improve base model capabilities. The new code & results are
GitHub - lyang36/IMO25: An AI agent system for solving International Mathematical Olympiad (IMO)...
From github.com
130K
Lin Yang
@lyang36
Jul 22, 2025
🚨 Olympiad math + AI: We ran Google’s Gemini 2.5 Pro on the fresh IMO 2025 problems. With careful prompting and pipeline design, it solved 5 out of 6 — remarkable for tasks demanding deep insight and creativity. The model could win gold! 🥇 #AI #Math #LLMs #IMO2025
349K
Lin Yang
@lyang36
Jul 1, 2025
I’m now officially an Associate Professor with tenure at UCLA, in the Electrical & Computer Engineering and Computer Science departments. Deeply grateful to my students, mentors, collaborators, and colleagues for their support along the way. @UCLAengineering
19K
Lin Yang
@lyang36
Jul 24, 2025
Code release! 🚀 Following up on our IMO 2025 results with the public LLM Gemini 2.5 Pro — here’s the full pipeline & general (non-problem-specific) prompts. 👉 [github.com/lyang36/IMO25] Have fun exploring! #AI #Math #LLMs #IMO2025
Lin Yang
@lyang36
Jul 22, 2025
🚨 Olympiad math + AI: We ran Google’s Gemini 2.5 Pro on the fresh IMO 2025 problems. With careful prompting and pipeline design, it solved 5 out of 6 — remarkable for tasks demanding deep insight and creativity. The model could win gold! 🥇 #AI #Math #LLMs #IMO2025
GitHub - lyang36/IMO25: An AI agent system for solving International Mathematical Olympiad (IMO)...
From github.com
47K
Lin Yang
@lyang36
Jul 22, 2025
Replying to @lyang36
github.com/lyang36/IMO25/… @huangyc07
IMO25/IMO25.pdf at main · lyang36/IMO25
From github.com
20K
Lin Yang
@lyang36
Jul 22, 2025
Replying to @jasondeanlee
It would be non-trivial for non-math person to achieve the same score. We have spent some time to carefully check the solutions. Regardless, the prompts are very general and can be applied to other models. We will release an automatic agent soon.
7.7K
Lin Yang
@lyang36
Jul 22, 2025
Replying to @lyang36
arxiv.org
Winning Gold at IMO 2025 with a Model-Agnostic...
The International Mathematical Olympiad (IMO) is widely regarded as the world championship of high-school mathematics. IMO problems are renowned for their difficulty and novelty, demanding deep...
13K
Lin Yang
@lyang36
Dec 14, 2024
Replying to @KevinKaichuang
“I hope this is an exception”
5.1K
Lin Yang
@lyang36
Jul 24, 2025
Replying to @lyang36
Compared the first version in our paper, this code removes problem specific hints completely. It just works!
6.8K
Lin Yang
@lyang36
Jul 22, 2025
Replying to @lyang36
There are already powerful models out there — but unlocking their full potential requires the right approach.
12K
Lin Yang
@lyang36
Nov 20, 2020
My postdoc mentor was Prof. Mengdi Wang at Princeton. She was super nice. She helped me a lot and inspired me to think out of the box. She is pivotal to my career path.
You’re unable to view this Post because this account owner limits who can view their Posts. Learn more
Lin Yang
@lyang36
Jul 16, 2020
Another two ICML posters on 07/16: Provably efficient RL with Linear Features (7a-7:45a, 6p-6:45p PDT), w/ @mengdiwang Provable Model-based RL using value-target regression: (7a-7:45a, 7p-7:45p PDT), w/ @AlexAyoub@CsabaSzepesvari @ZeyuJia, @mengdiwang (1/2)
Lin Yang
@lyang36
Aug 27, 2025
Replying to @AtaeiMe
It appears this model does not follow the prompt instructions. I did not figure out whether it’s my deployment problem or the model is not strong enough.
4.8K
Lin Yang
@lyang36
Dec 14, 2024
Replying to @NeurIPSConf
This is clearly an explicit racist comment against Chinese instead of implicit bias.
3.2K