It goes beyond the title with direct discussion of like, yeah, it's, including: We barely covered the news yesterday.
Key takeaways
01Yesterday, Amir over at The Information reported that OpenAI is quietly, quote, quietly using loop transformers that don't show the model's thinking when scaled up, but they make t
02So neuralese for context is like the
03The thing that I want to understand is like usually these models are trained so that they're changing weights into text at some point
Best for
listeners looking for a practical AI episode debrief