Bayes Estimators: Loss functions with flexible signature

stefanradev93 / BayesFlow

A Python library for amortized Bayesian workflows using generative neural networks.

https://bayesflow.org/

MIT License

286 stars 45 forks source link

Bayes Estimators: Loss functions with flexible signature #147

Closed han-ol closed 4 months ago

han-ol commented 4 months ago

Implementation proposal to the signature change of the loss_fun used for Amortized Bayes Estimators as discussed in https://github.com/stefanradev93/BayesFlow/issues/121.

It stays as simple as before to use norm_ord, but the optional loss_fun argument expects a function object with two arguments, rather than the difference of parameters and network output.

stefanradev93 commented 4 months ago

Great idea!

han-ol commented 4 months ago

Thanks!