wf_psf.utils.optimizer

Optimizer utilities for WF-PSF.

This module provides utility functions to create optimizers for training or evaluation of PSF models.

Author:

Jennifer Pollack <jennifer.pollack@cea.fr>

Functions

get_optimizer([optimizer_config])

Return a compiled optimizer instance based on configuration or name.

is_optimizer_instance(obj)

wf_psf.utils.optimizer.get_optimizer(optimizer_config=None, **overrides)[source]

Return a compiled optimizer instance based on configuration or name.

Parameters:

  • optimizer_config (RecursiveNamespace, dict, or str, optional) – Optimizer configuration (from YAML or programmatically), or string name.

  • **overrides (keyword arguments) – Optional hyperparameters to override values in optimizer_config (e.g., learning_rate, beta_1, beta_2, epsilon, amsgrad).

Return type:

tf.keras.optimizers.Optimizer

wf_psf.utils.optimizer.is_optimizer_instance(obj)[source]