--model_path

Description

Sets the model file directory, which can store multiple model files. In tuning for PyTorch training, the model file refers to the dumped operator graph. For details about the dump operation, see Dumping the Operator Graph.

Model files in this path can be tuned in batches.

See Also

• During tuning, --model or -m or --model_path must be specified.
• This option is mutually exclusive with --output_type or --out_nodes.

Arguments

Argument: Model file directory.

Format: The directory can contain letters, digits, underscores (_), hyphens (-), periods (.), and Chinese characters.

Suggestions and Benefits

None

Examples

--model_path=${HOME}/test/

Dependencies and Restrictions

Models in model_path must use the same tuning options.

During offline tuning of TensorFlow-based training, the argument of this option cannot be the same as the current working directory of AOE.