Overview

This section describes the options used in the CLI scenario.

If options queried with the amct_caffe calibration --help command are not described in Table 1, they are reserved or applicable to other SoC versions. You do not need to pay attention to such options.
The amct_onnx command can be organized in either of the following ways:
- amct_caffe calibration param1=value1 param2=value2 ... (No space is allowed before value. Otherwise, it will be truncated, and the value of param is empty.)
- amct_xxx calibration param1 value1 param2 value2 ...

Replace xxx with the actual framework name, for example, Caffe, TensorFlow, or ONNX.

**Table 1** Command-line options
Option	Command-Line Options	Use framework/network	Application Scenarios
--help or --h	Displays help information.	Caffe TensorFlow ONNX	Post-training quantization: optional QAT Model Adaptation to CANN Model: Optional
--model	Sets the path (including the file name) of the model file to be quantized. Caffe: The file is in .prototxt format. TensorFlow: The file is in .pb format. ONNX: The file is in .onnx format.	Caffe TensorFlow ONNX	Post-training quantization: mandatory (Mandatory) Adapting QAT Models to CANN Models
--weights	Sets the directory (including the file name) of the weight file (.caffemodel) to be quantized. This option applies only to Caffe models.	Caffe	Post-training quantization: mandatory
--outputs	Sets the name of the output tensor for the source model.	TensorFlow	Post-training quantization: mandatory (Mandatory) Adapting QAT Models to CANN Models
--save_path	Sets the model save path.	Caffe TensorFlow ONNX	Post-training quantization: mandatory (Mandatory) Adapting QAT Models to CANN Models
--input_shape	Sets the shape of the model input. The shape must be the same as that of the input data after dataset processing. Required if --evaluator is not included.	Caffe TensorFlow ONNX	Post-training quantization: optional QAT model adaptation to CANN model: not applicable
--batch_num	Sets the number of batches for the inference stage of PTQ.	Caffe TensorFlow ONNX	Post-training quantization: optional QAT model adaptation to CANN model: not applicable
--calibration_config	Sets the path (including the file name) of the simplified configuration file for PTQ.	Caffe TensorFlow ONNX	Post-training quantization: optional QAT model adaptation to CANN model: not applicable
--data_dir	Sets the path of the .bin dataset that matches the model. Required if --evaluator is not included.	Caffe TensorFlow ONNX	Post-training quantization: optional QAT model adaptation to CANN model: not applicable
--data_types	Sets the input data type. Required if --evaluator is not included.	Caffe TensorFlow ONNX	Post-training quantization: optional QAT model adaptation to CANN model: not applicable
--evaluator	Sets the Python script that is based on the Evaluator base class and contains the evaluator. This option is mutually exclusive with --input_shape, --data_dir, and --data_types.	Caffe TensorFlow ONNX	Post-training quantization: optional QAT model adaptation to CANN model: not applicable
--gpu_id	Specifies a GPU for quantization.	Caffe	Post-training quantization: optional

Parent topic: Command-line options