danielward27 / flowjax

https://danielward27.github.io/flowjax/
MIT License
82 stars 10 forks source link

Option to control returning best parameters vs last parameters in training #137

Closed danielward27 closed 7 months ago

danielward27 commented 7 months ago

Adds an option to allow returning the last model, rather than the one which led to the minimum loss. Useful e.g. if loss is high variance to ignore spurious low loss values, or for debugging a model that starts returning nans during training.