--is_input_adjust_hw_layout

Applicability

Product

Supported (√/x)

Atlas A3 training product/Atlas A3 inference product

Atlas A2 training product/Atlas A2 inference product

Atlas 200I/500 A2 inference product

Atlas inference product

Atlas training product

Description

Sets the data type and format of the network inputs to FP16 and NC1HWC0, respectively.

See Also

  • Use this option in conjunction with --input_fp16_nodes. If this option is set to true, the data type and format of the --input_fp16_nodes inputs are set to FP16 and NC1HWC0, respectively.
  • This option cannot be used when --framework or -f is set to 1 to use the MindSpore network model.

Arguments

Arguments: false or true

Default: false

Suggestions and Benefits

None

Examples

When there is only one input_fp16_nodes, the usage example is provided below:

--framework=3 --model=${HOME}/module/resnet50_tensorflow_1.7.pb --output=${HOME}/test/out/tf_resnet50 --is_input_adjust_hw_layout=true  --input_fp16_nodes="data"

When there are two or more input_fp16_nodes, the usage example involving two such nodes is provided below:

--framework=3 --model=${HOME}/module/resnet50_tensorflow_1.7.pb --output=${HOME}/test/out/tf_resnet50 --is_input_adjust_hw_layout=true,true  --input_fp16_nodes="data1;data2"

Dependencies and Restrictions

None