AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
Is There a Controllability of Language Models?
We want to build a domain general platform. How do you make it so that the models that you have are truly generalizable? Ray, this is has been an ongoing line of work for in deep learning for many years. I think the last two years or so has been phenomenal in terms of how much more flexible and generalizable these models are. But still not ther there is a very, very long way to go. The other thing is a data ray, these models are tremely data hungry. And i see a lot of value in someone like myself having spent time thinking about problems that like, a pure deep learning person would not have had,. It kind of brings to building a