<br />

|---|
| [![](https://www.tensorflow.org/images/GitHub-Mark-32px.png) View source on GitHub](https://github.com/tensorflow/models/blob/v2.20.0/official/modeling/optimization/optimizer_factory.py) |

Optimizer factory class.

## Classes

[`class OptimizerFactory`](https://www.tensorflow.org/api_docs/python/tfm/optimization/OptimizerFactory): Optimizer factory class.

## Functions

[`register_optimizer_cls(...)`](https://www.tensorflow.org/api_docs/python/tfm/optimization/register_optimizer_cls): Register customize optimizer cls.

<br />

<br />

<br />

| ## Other Members ||
|---|---|
| LEGACY_OPTIMIZERS_CLS | <br /> { 'adafactor': 'Unimplemented', 'adafactor_keras': <class 'tf_keras.src.optimizers.adafactor.Adafactor'>, 'adagrad': <class 'tf_keras.src.optimizers.legacy.adagrad.Adagrad'>, 'adam': <class 'tf_keras.src.optimizers.legacy.adam.Adam'>, 'adam_experimental': <class 'tf_keras.src.optimizers.adam.Adam'>, 'adamw': <class 'official.modeling.optimization.legacy_adamw.AdamWeightDecay'>, 'adamw_experimental': <class 'tf_keras.src.optimizers.adamw.AdamW'>, 'lamb': <class 'official.modeling.optimization.lamb.LAMB'>, 'lars': <class 'official.modeling.optimization.lars.LARS'>, 'rmsprop': <class 'tf_keras.src.optimizers.legacy.rmsprop.RMSprop'>, 'sgd': <class 'tf_keras.src.optimizers.legacy.gradient_descent.SGD'>, 'sgd_experimental': <class 'tf_keras.src.optimizers.sgd.SGD'>, 'slide': 'Unimplemented' } <br /> |
| LR_CLS | <br /> { 'cosine': <class 'official.modeling.optimization.lr_schedule.CosineDecayWithOffset'>, 'exponential': <class 'official.modeling.optimization.lr_schedule.ExponentialDecayWithOffset'>, 'polynomial': <class 'official.modeling.optimization.lr_schedule.PolynomialDecayWithOffset'>, 'power': <class 'official.modeling.optimization.lr_schedule.DirectPowerDecay'>, 'power_linear': <class 'official.modeling.optimization.lr_schedule.PowerAndLinearDecay'>, 'power_with_offset': <class 'official.modeling.optimization.lr_schedule.PowerDecayWithOffset'>, 'step_cosine_with_offset': <class 'official.modeling.optimization.lr_schedule.StepCosineDecayWithOffset'>, 'stepwise': <class 'official.modeling.optimization.lr_schedule.PiecewiseConstantDecayWithOffset'> } <br /> |
| NEW_OPTIMIZERS_CLS | <br /> { 'adafactor': 'Unimplemented', 'adafactor_keras': <class 'tf_keras.src.optimizers.adafactor.Adafactor'>, 'adagrad': <class 'tf_keras.src.optimizers.adagrad.Adagrad'>, 'adam': <class 'tf_keras.src.optimizers.adam.Adam'>, 'adam_experimental': <class 'tf_keras.src.optimizers.adam.Adam'>, 'adamw': <class 'official.modeling.optimization.legacy_adamw.AdamWeightDecay'>, 'adamw_experimental': <class 'tf_keras.src.optimizers.adamw.AdamW'>, 'lamb': <class 'official.modeling.optimization.lamb.LAMB'>, 'lars': <class 'official.modeling.optimization.lars.LARS'>, 'rmsprop': <class 'tf_keras.src.optimizers.rmsprop.RMSprop'>, 'sgd': <class 'tf_keras.src.optimizers.sgd.SGD'>, 'sgd_experimental': <class 'tf_keras.src.optimizers.sgd.SGD'>, 'slide': 'Unimplemented' } <br /> |
| SHARED_OPTIMIZERS | <br /> { 'adafactor': 'Unimplemented', 'adafactor_keras': <class 'tf_keras.src.optimizers.adafactor.Adafactor'>, 'adam_experimental': <class 'tf_keras.src.optimizers.adam.Adam'>, 'adamw': <class 'official.modeling.optimization.legacy_adamw.AdamWeightDecay'>, 'adamw_experimental': <class 'tf_keras.src.optimizers.adamw.AdamW'>, 'lamb': <class 'official.modeling.optimization.lamb.LAMB'>, 'lars': <class 'official.modeling.optimization.lars.LARS'>, 'sgd_experimental': <class 'tf_keras.src.optimizers.sgd.SGD'>, 'slide': 'Unimplemented' } <br /> |
| WARMUP_CLS | <br /> { 'linear': <class 'official.modeling.optimization.lr_schedule.LinearWarmup'>, 'polynomial': <class 'official.modeling.optimization.lr_schedule.PolynomialWarmUp'> } <br /> |

<br />