Overview

--help or --h

Displays help information.

Optional

N/A

--mode

Sets the work mode.

Optional

0

--model

Sets the model file directory, including the file name.

Required

N/A

--weight

Sets the weight file directory, including the file name.

Optional

N/A

--om

Sets the directory (including the file name) of the offline model or original model file to be converted to JSON format.

Optional

N/A

--framework

Sets the framework of the original model.

Required

N/A

--input_format

Sets the input data format.

Optional

NCHW (Caffe)

NHWC (TensorFlow)

--input_shape

Sets the input shape.

Optional

N/A

--dynamic_batch_size

Sets dynamic batch size choices. Applies to the scenario where image count per inference batch is unfixed.

Optional

N/A

--dynamic_image_size

Sets dynamic image size choices. Applies to the scenario where the resolution of images input for inference is unfixed.

Optional

N/A

--dynamic_dims

Sets dynamic dimension size choices in ND format. Applies to the scenario where the dimension size for inference is unfixed.

Optional

N/A

--singleop

Sets the single-operator definition file, used to convert a single-operator JSON file into an offline model adapted to the Ascend AI Processor.

Optional

N/A

--output

• Sets the directory (including the file name) of the offline model converted from a network from open-source framework.
• Sets the directory of the generated single-operator model converted from a single-operator JSON file.

Required

N/A

--output_type

Sets the output data type of a network or an output node.

Optional

N/A

--check_report

Sets the precheck result file directory, including the file name.

Optional

check_result.json

--json

Sets the directory (including the file name) of the JSON file converted from the offline model or original model file.

Optional

N/A

--soc_version

Sets the target SoC version.

Required

N/A

--aicore_num

Sets the number of AI Cores for model build.

Optional

2

--out_nodes

Sets the output nodes.

Optional

N/A

--input_fp16_nodes

Sets the name of the FP16 input node.

Optional

N/A

--insert_op_conf

Sets the configuration file directory (including the file name) of an operator to be inserted, for example, the aipp operator for data preprocessing.

Optional

N/A

--op_name_map

Sets the directory (including the file name) of the mapping configuration file of a custom operator. The function of a custom operator varies according to the network. You can specify the mapping between the custom operator and the actual custom operator running on the network.

Optional

N/A

--is_input_adjust_hw_layout

Sets the data type and format of the network inputs to fp16 and NC1HWC0, respectively.

Optional

false

--is_output_adjust_hw_layout

Sets the data type and format of the network outputs to fp16 and NC1HWC0, respectively.

Optional

false

--disable_reuse_memory

Enables memory reuse.

Optional

0

--fusion_switch_file

Sets the fusion switch configuration file directory, including the file name.

Optional

N/A

--enable_scope_fusion_passes

Enables specific fusion rules in build.

Optional

N/A

--enable_small_channel

Enables small channel optimization. If enabled, performance benefits are generated at the first convolutional layer with channel <= 4.

Optional

0

--buffer_optimize

Enables buffer optimization.

Optional

l2_optimize

--precision_mode

Selects the operator precision mode.

Optional

force_fp16

--auto_tune_mode

Selects the operator auto tuning mode.

Optional

N/A

--op_select_implmode

Selects the operator implementation mode.

Optional

high_performance

--optypelist_for_implmode

Lists the operator types. The listed operators use the mode specified by the --op_select_implmode option.

Optional

N/A

--op_debug_level

Enables TBE operator debug during operator build.

Optional

0

--dump_mode

Dumps a JSON file with shape information.

Optional

0

--log

Sets the level of logs to be printed during ATC model conversion.

Optional

null