Lilian Weng

new post on harness engineering for AI self-improvement: t.co/ZYvGfVs61k It is hard to forecast how much the future of RSI will rely on harnesses. Likely harness engineering will evolve in the direction of self-improvement and enable auto-research, and, in turn, smarter models keeps harnesses simple. Even when many harness improvement get eventually internalized into core model, the need to specify goals and context will not disappear.

A super long overdue (3+ years?) post on scaling laws. Compute is expensive. Scaling laws are a way to help us reason about the optimal compute allocation between data and model size before committing to a large run. The post covers what scaling laws predict, how compute-optimal allocation works, why Kaplan et al. and Chinchilla disagree, and how data limits + fitting details make extrapolation tricky. t.co/HP26eJvjHB

I wonder how many more people will just ask model to summarize the post, in comparison to a year ago. 😂

Soon I should set up the model to update Lil’Log automatically but I’m not there yet.

We would love to see more collaboration and research in the field of human-AI interactivity. Check it out! x.com/thinkymachines…

I only recently read more about the concept of system accidents by Charles Perrow, very insightful and relatable.

In the past few months, we had a lot of fun (and stress 😅) to produce 12 versions (+ many subversions) and 137 pages in our training run log book. Turns out human-human collaboration is important to improving human-AI collaboration. 😊 x.com/thinkymachines…

Building technologies for better human-AI collaboration on next gen hardware at scale. Exciting. x.com/thinkymachines…

I’ve been telling people this a lot today: I enjoy so much working with people who care about what they are building and craftsmanship. It is a privilege to have a chance to work on something I’m passionate about, beyond making a living. I cherish it and don’t take it for

On-policy distillation provides an elegant way to use the teacher model as a process reward model to provide dense reward while preventing SFT style "OOD shock" during rollout. x.com/thinkymachines…

GPUs are expensive and setting up the infrastructure to make GPUs work for you properly is complex, making experimentation on cutting-edge models challenging for researchers and ML practitioners. Providing high quality research tooling is one of the most effective ways to