--is_output_adjust_hw_layout

Applicability

Product

Supported

Atlas A3 training products/Atlas A3 inference products

Atlas A2 training products/Atlas A2 inference products

Atlas 200I/500 A2 inference products

Atlas inference products

Atlas training products

Description

Sets the network model output data type to float16 and data format to NC1HWC0 together with --out_nodes. Setting this option separately does not take effect, but the model will be successfully converted.

See Also

  • If --is_output_adjust_hw_layout is set to true, it needs to be used together with --out_nodes. For nodes specified by --out_nodes, the output data type is float16 and output data format is NC1HWC0.

    The number of values of --is_output_adjust_hw_layout must match the number of nodes specified by --out_nodes. Use commas (,) to separate multiple values.

  • When --framework is set to 1, this option is invalid in the MindSpore framework, but the model will be successfully converted.

Argument

  • true: Sets the output data type to float16 and output data format to NC1HWC0 for a network model.
  • false (default): not specified.

Suggestions and Benefits

None

Example

  • If one node is configured by --out_nodes:
    atc --model=$HOME/module/resnet50_tensorflow*.pb --framework=3 --output=$HOME/module/out/tf_resnet50 --is_output_adjust_hw_layout=true  --out_nodes="prob:0" --soc_version=<soc_version>
  • If multiple nodes are configured by --out_nodes:
    atc --model=$HOME/module/resnet50_tensorflow*.pb --framework=3 --output=$HOME/module/out/tf_resnet50 --is_output_adjust_hw_layout="true,true"  --out_nodes="prob:0;prob:1" --soc_version=<soc_version>

Dependencies and Restrictions

None