search

aleximmer / Laplace

Laplace approximations for Deep Learning.
https://aleximmer.github.io/Laplace
MIT License
436 stars 63 forks source link

Make last-layer Jacobians agnostic to NN output shape #112

Closed runame closed 1 year ago

runame commented 1 year ago

Defines output_size such that it is agnostic to the specific shape of the neural network output f.