--feature

Description

This option applies only to TensorFlow models.

Enables model compression features.

See Also

None

Argument

Argument:

• ptq (default): enables only PTQ.
• approximate: enables only calibrated approximation of the operator.
• approximate|ptq: enables both the calibrated approximation of the operator and PTQ.

Format: If both the calibrated approximation of the operator and PTQ of the model are enabled, the argument must be enclosed in double quotation marks ("") (for example, "approximate|ptq").

Suggestions and Benefits

None

Example

--feature="approximate|ptq"