It turns out multi step backpropaganda is better.
paper has a beautiful way of improving backpropagation. One iteration cleanly gets us backprop, multiple iterations get us a preconditioned update.
Key is that direction has become much more important. Being close to the “source”, having the right directions gets you very far. It also changes what an organization that does research looks like.
Mark Chen and John Schulman on what today's AI research interns can do, and where they still need human direction:
Mark: "I do believe that we've built AI research interns today."
"This is most often deployed in places where you give the model something fairly scoped to do,
Very confused, how can you solve millennium prize problem and inspire us including me but cant get models to behave.
Does this imply that mathematicians are recluse meme and dont follow instructions
I will not be writing a full review of GPT-6.2 Astra / Sol.
I had a terrible experience testing them. They were essentially unusable.
I do not recommend these models.
Some of you have been asking very politely. Some of you have just been asking.
GPT-6.2 Astra and Sol are now available.
Super efficient and built to perform where it counts—the real world.
openai.com/index/introduc…