Ming Yin
@MingYin_0312
ML, RL, AI. @Princeton Postdoc. PhDs in CS & STATs. Ex @awscloud AI. undergrad @USTC Math. Area Chair @NeurIPS @ICML.
Really excited to work with @AndrewYNg and @DeepLearningAI on this new course on post-training of LLMs—one of the most creative and fast-moving areas in LLM development. We cover the key techniques that turn pre-trained models into helpful assistants: SFT, DPO, and online RL.…
New Course: Post-training of LLMs Learn to post-train and customize an LLM in this short course, taught by @BanghuaZ, Assistant Professor at the University of Washington @UW, and co-founder of @NexusflowX. Training an LLM to follow instructions or answer questions has two key…
🤔 How do you get an LLM to reason like a CRISPR pro—or any top scientist? By training it on real expert conversations. 🛠️ What we built • An automated pipeline that distills learning signals from 10 + years of genomics discussions • Genome-Bench: 3,000 + curated Q&As on…
🤔 How do you train an AI model to think and reason like a biology expert? We found the answer: let it learn from real expert discussions! Checkout our recent work on a breakthrough approach to improve LLM scientific reasoning - by learning directly from 10+ years of genomics…
“There are no authorities in science,” says Turing Award winner @RichardSSutton, Amii Fellow & Canada @CIFAR_News AI Chair. Sit down with Rich and @camlinke as they discuss the journey to this moment. Watch now: hubs.la/Q039xBP-0 #TuringAward #AI #ReinforcementLearning
Princeton has launched AI for Accelerating Invention. Led by professors @MengdiWang10 and @ryan_p_adams, the AI^2 initiative aims to achieve faster breakthroughs across engineering disciplines: bit.ly/4fywVWc
I feel more people should have seen this. It reminds me of David Sliver’s talk at RLC on ‘Towards Superhuman Intelligence’ tinyurl.com/rlcds. I believe RL will be the key for brining fundamental improvement over existing (Gen)AI.
2022: I never wrote a RL paper or worked with a RL researcher. I didn’t think RL was crucial for AGI Now: I think about RL every day. My code is optimized for RL. The data I create is designed just for RL. I even view life through the lens of RL Crazy how quickly life changes
Finally, after two years, our paper is out @NatComputSci ! Over the past two years, the community has witnessed rapid growth from distribution learning and substructure design to the optimization and fine-tuning of generative models for designing various types of molecules!
🚨@rneschneuing, @charlieharris01, @YuanqiD, @mmbronstein, @befcorreia and colleagues evaluate how diffusion models can be used to address structure-based drug design problems. nature.com/articles/s4358…