radarsat1 on State-space models can learn in-context by gradient descent · Hackerly

Comment on State-space models can learn in-context by gradient descent parent

Converging slowly on Kalman filters, calling it now.