Patience for early stopping

htjb / margarine

Code to replicate posterior probability distributions with bijectors/KDEs and perform marginal KL/bayesian dimensionality calculations.

MIT License

13 stars 8 forks source link

Patience for early stopping #55

Open htjb opened 6 months ago

htjb commented 6 months ago

Currently, the patience for early stopping is set to 2% of the requested epochs. This is usually fine for a sufficiently large number of requested epochs, but in practice the user should be able to specify how long they want to wait with a kwarg called patience. Can default to 2%.