Lost in the Middle: How Language Models Use Long Contexts
Deep Papers
00:00
The Future of Machine Learning
It's fun to think about how we are producing the same as these transformer architectures of, you know, being able to retrieve documents in a similar way. So I feel like it's encouraging like maybe we're on the right track, but we have to figure out better ways to work within these systems. Maybe we won't see a perfect world where the model will be able to grab context from every bit of its input. But I think it's going to be a way that we can adapt to understanding like, okay, we need to put that at the beginning of the end. What does that mean for shortening context? And this makes me want to ask a lot of questions and dive
Transcript
Play full episode
Remember Everything You Learn from Podcasts
Save insights instantly, chat with episodes, and build lasting knowledge - all powered by AI.
