🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

Congrats! Gregory Lau Apivich H. (Kaotoo) He Zhenfeng Sebastian Tay Wu Zhaoxuan Rui Qiao qphong ruth Dai Zhongxiang @ZCODE patrick jaillet Accepted ICLR 2026 #ICLR2024: #ShapleyValue #DataValuation #FederatedLearning #BayesianOptimization #NeuralArchitectureSearch #NeuralTangentKernel #AI4Science

Looking forward to chatting about our work, PINNACLE! Later today at #ICLR2024 Poster Session 6 May 9 Thu 4:30PM Halle B #30 openreview.net/forum?id=GzNaC…

Visit the poster of Gregory Lau Xinyuan Dao Hieu Chen Jiangwei chuan-sheng on the Waterfall framework for robust & scalable text #Watermarking to protect IP/copyrights of text ICML Conference #ICML2024 Workshop on Foundation Models in the Wild (26 Jul, icml-fm-wild.github.io) #LLMs.

We'll give an oral presentation on PINNACLE 🏔️ at ICML Conference #ICML2024 Workshop on AI for Science, which is based on #PINNs #NeuralTangentKernel to jointly optimize the selection of different training points via #ActiveLearning (26 Jul, ai4sciencecommunity.github.io/icml24.html)! #AI4Science #PDEs

How can we solve #InverseProblems with limited data & compute for simulations, esp. in science/engineering settings? PIED 🥧 is an #ExperimentalDesign framework for #PDE-based inverse problems that efficiently optimizes for the best sensor placements (1/n). ICML Conference #ICML2024

ICML Conference Gregory Lau Apivich H. (Kaotoo) PINNACLE 🏔️ won the best paper award (out of 225 submissions) at ICML Conference #ICML2024 Workshop on AI for Science! Congrats to Gregory Lau and Apivich H. (Kaotoo)!!!

The #EMNLP2024 EMNLP 2025 (main) work of Gregory Lau Xinyuan Dao Hieu Chen Jiangwei chuan-sheng proposes the Waterfall framework for robust & scalable text #Watermarking to protect IP/copyrights of text for #LLM data provenance. 1st NLP conf paper! arxiv.org/abs/2407.04411

🔬 Deepmind's recent Nature paper on #watermarking #LLM outputs highlights how watermarking methods are increasingly important in the era of #LLMs (1/n)

1st NLP conference for GLOW.AI EMNLP 2025 #EMNLP2024: Gregory Lau Xinyuan! How different would it be from ICML Conference #ICML2024 NeurIPS Conference #NeurIPS2024? Better food? 😋 Interested to apply for faculty or PhD in AI NUS Computing, ping me on Whova to meet up.

Gregory Lau Xinyuan presenting Waterfall framework for text #Watermarking to protect IP/copyrights of text for #LLM data provenance EMNLP 2025 #EMNLP2024!!!

Xinyuan and Gregory Lau presenting our position paper on data-centric AI in the age of #LLM EMNLP 2025 #EMNLP2024! Thanks to Xinyi Xu for preparing the poster! Joint work with Pang Wei Koh.

GLOW.AI (comp.nus.edu.sg/~lowkh) will be NeurIPS Conference #NeurIPS2024: Hu.Wenyang Wu Zhaoxuan Zijian Zhou Zhuanghua Liu Rui Qiao Gregory Lau. Interested to apply for faculty or PhD in AI NUS Computing, ping me on Whova to meet up. See you soon!

Given a single model, how do we improve an #LLM’s reasoning performance with limited resources 💻 and inference time⌛️? Can a smaller 1.5B model outperform a 7B model without incurring long inference time from sequential queries? (1/n) NeurIPS Conference #NeurIPS2024 #LLMs

NeurIPS Conference Introducing Dipper: a framework to create #LLMs ensembles from an optimized set of diverse reasoning prompts to improve perf. Unlike sequential inference time methods, Dipper runs queries in parallel, making it super fast ⏩ and effective. (2/n) NeurIPS Conference #NeurIPS2024

NeurIPS Conference Furthermore, Dipper can work with LLM APIs without model access 📦! (3/n) NeurIPS Conference #NeurIPS2024

NeurIPS Conference We showed how a small ensemble of 3 1.5B models can outperform a 7B model on MATH with almost the same inference time & <3x compute for a normal query due to accelerated batch inference methods 😱 ! C u NeurIPS Conference #NeurIPS2024 Workshop on Foundation Model Interventions (n/n)

🧪 A key challenge to solving #InverseProblems, especially in science and engineering settings, is limited data and compute for simulations. 🥧 The ICLR 2026 work of Apivich H. (Kaotoo) Gregory Lau , PIED, is an #ExperimentalDesign framework for #PDE inverse problems that efficiently

ICLR 2026 Apivich H. (Kaotoo) Gregory Lau Gregory Lau and Apivich H. (Kaotoo) are getting PIED ICLR 2026 #ICLR2025 🦶🥧🤣

Introducing WaterDrum🛢️, the first data-centric #LLM #unlearning metric that leverages robust text #watermarking💧to provide an effective, practical, and resilient way to evaluate LLM unlearning performance😉! (1/n) #MachineUnlearning #LLMs

Excited to be visiting University of Oxford at Oxford Statistics @OxCSML over the summer, hosted by Desi R. Ivanova! Just arrived, and already having many interesting discussions. Looking forward to meeting people, exchanging ideas and engaging in research collaborations!