Kangwook Lee
@Kangwook_Lee
UW Madison / KRAFTON AI
🚨New Paper Alert🚨 Could generative agents powered by LLMs transform social science by accurately simulating human social behaviors at scale? We tested this possibility with virtual humans facing disease threats in "Infected Smallville." 🧵
SK텔레콤에서 독자적으로 구축한 LLM 모델 'A.X(에이닷 엑스) 3.1 라이트'를 공개했습니다. 우수한 한국어 처리능력은 물론, 경량화된 성능과 높은 효율을 바탕으로 다양한 사양의 모바일 기기에서 최적의 서비스 품질을 구현할 수 있습니다. ▶ bit.ly/4kyBYs6 #SK텔레콤 #에이닷엑스 #AI
If you're at ICML, check out our posters 😄! 11 am: @nayoung_nylee will talk about “self improvment" techniques to solve length generalization! 4:30 pm: Kevin, Wonjun and @yzeng58 will present theoretical/empirical analysis of PEFT methods for SSMs & a new method!
Looking forward to seeing folks tomorrow afternoon!
Working in active learning, I often hear a common complaint — active learning doesn’t work sometimes. Our ICML 2025 paper JUST WORKS* when you have class imbalance. @rdnowak will present it Tuesday afternoon. Short 🧵below Paper: arxiv.org/abs/2312.09196
New lecture recordings on RL+LLM! 📺 This spring, I gave a lecture series titled **Reinforcement Learning of Large Language Models**. I have decided to re-record these lectures and share them on YouTube. (1/7)