Blog post: Distributed cloud builds for everyone
Ranger update! He turned 6 months old about a week ago. Here he is celebrating Memorial Day yesterday with his very first slice of watermelon, which he looooooved.
Blog post: Distributed cloud builds for everyone
I published another blog post about Llama, this time trying to outline the philosophy behind its design decisions, and why I'm excited about the prospect of making distributed cloud builds much more broadly accessible. Now that I've "released" Llama I'm probably going to be doing more publishing about it on the blog and/or on a future Llama website, to try to get that vision out more broadly.
Life update
I've joined Anthropic AI! I've actually been working there for a month or two, but they just launched on Friday so now I have something to link to.
I’ll be mostly working with Chris Olah and Catherine Olsson on understanding what the heck is actually going on inside ML models; You can check out Chris et al's Circuits threads on distill.pub for a sense of what we'll be doing.
I'm really excited by Anthropic's dedication to treating AI/ML as a systematic empirical science, and building a systematic understanding both of individual models and of designing and training models more broadly.
For me, it really meshes with my philosophy of wanting to deeply understand how computer systems work. I’ve written previously about how deep learning systems are a black box that can't be understood in detail; I'm excited to see if we can prove me wrong.
Working full time definitely leaves less time and energy for blogging and side projects like Llama, so don't be surprised if it's a bit quieter over here. I do intend to continue writing and publishing when I can, though.