convert_qat_model

Applicability

Product	Supported
Atlas 350 Accelerator Card	√
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

Converts a TensorFlow QAT model to a quantized model that serves for both accuracy simulation on the CPU and GPU and deployment on the AI processor.

Prototype

convert_qat_model(pb_model, outputs, save_path, record_file=None)

Parameters

Parameter	Input/Output	Description
pb_model	Input	Path of the QAT model to be adapted. A string.
outputs	Input	List of output operators in a graph. A list.
record_file	Input	Path of the quantization factor record file (.txt) computed by the user. A string. Default: None
save_path	Input	Model save path. Must include the prefix of the model name, for example, **./quantized_model/model***. A string.

Returns

Returns a list of output operators in a graph.

Restrictions

Only TensorFlow models that contain the FakeQuantWithMinMaxVars and FakeQuantWithMinMaxVarsPerchannel operators are supported. The model format is .pb.

Example

import amct_tensorflow as amct
convert_qat_model(pb_model, outputs, save_path)

Flush file: a .pb model file that can be used for accuracy simulation in the TensorFlow environment or offline inference on the AI processor.

When adaptation is performed again, the preceding files output by the API will be overwritten.

Parent topic: Model Adaptation APIs