tf.data.experimental.AutotuneAlgorithm

Represents the type of autotuning algorithm to use.

Compat aliases for migration

See Migration guide for more details.

DEFAULT: The default behavior is implementation specific and may change over time.

HILL_CLIMB: In each optimization step, this algorithm chooses the optimial parameter and increases its value by 1.

GRADIENT_DESCENT: In each optimization step, this algorithm updates the parameter values in the optimal direction.

MAX_PARALLELISM: Similar to HILL_CLIMB but uses a relaxed stopping condition, allowing the optimization to oversubscribe the CPU.

STAGE_BASED: In each optimization step, this algorithm chooses the worst bottleneck parameter and increases its value by 1.

Class Variables
DEFAULT	`<AutotuneAlgorithm.DEFAULT: 0>`
GRADIENT_DESCENT	`<AutotuneAlgorithm.GRADIENT_DESCENT: 2>`
HILL_CLIMB	`<AutotuneAlgorithm.HILL_CLIMB: 1>`
MAX_PARALLELISM	`<AutotuneAlgorithm.MAX_PARALLELISM: 3>`
STAGE_BASED	`<AutotuneAlgorithm.STAGE_BASED: 4>`

Except as otherwise noted, the content of this page is licensed under the Creative Commons Attribution 4.0 License, and code samples are licensed under the Apache 2.0 License. For details, see the Google Developers Site Policies. Java is a registered trademark of Oracle and/or its affiliates. Some content is licensed under the numpy license.

Last updated 2023-10-06 UTC.

Class Variables