Differentiable Programming: A Cool Approach to Problem Solving

The fusion models are a better example of this, but it's a very, so real quick, are you familiar with the idea of sort of differentiable programming or? No. You know, and there's a library in Jack says using a lot for stuff. Basically, you describe your problem in terms of the parameters that a program takes and the output you want. That's why it's called differentiable because you can take the derivative of a program rather than a function. And then you just do gradient descent and you get the optimal weights for that.