save_quant_retrain_model

Function Usage

Inserts operators such as AscendQuant and AscendDequant into the retrained model to generate a fake-quantized model for simulation and a deployable model for deployment.

Constraints

None

Prototype

save_quant_retrain_model(retrained_model_file, retrained_weights_file, save_type, save_path, scale_offset_record_file = None, config_file = None)

Command-Line Options

Option	Input/Return	Description	Restriction
retrained_model_file	Input	Definition file of the retrained Caffe model (.prototxt).	A string Restriction: For layers for inference, the settings in LayerParameter in retrained_model_file must meet inference requirements. For example, use_global_stats of the BatchNorm layer must be set to 1.
retrained_weights_file	Input	Weight file of the Caffe model (.caffemodel).	A string
save_type	Input	Type of the model to be saved. Fakequant indicates the storage precision simulation model. Deploy: a deployable model that can be used for inference on Ascend AI Processor. Both indicates that both models are stored.	A string
save_path	Input	Model save path. Must include the prefix of the model name, for example, *./quantized_model/model**.	A string
scale_offset_record_file	Input	Quantization factor record file.	Data type: string (Initialized to None by default.)
config_file	Input	Quantization configuration file.	Data type: string (Initialized to None by default.)

Return Value

None

Outputs

A fake-quantized model for accuracy simulation in the Caffe environment and its weight file, with names containing the fake_quant keyword.
A deployable model and its weight file, with names containing the deploy keyword. The model can be deployed on Ascend AI Processor after being converted by the ATC tool.

Examples

from amct_caffe import amct
retrained_model_file =  './pre_model/retrained_resnet50.prototxt'
retrained_weights_file = './pre_model/resnet50_solver_iter_35000.caffemodel'
scale_offset_record_file = './record.txt'
# Insert the API, save the retrained model as a .prototxt model file and a .caffemodel weight file, and generate four files in ./result: model_fake_quant_model.prototxt, model_fake_quant_weights.caffemodel, model_deploy_model.prototxt and model_deploy_weights.prototxt.
amct.save_quant_retrain_model(retrained_model_file, retrained_weights_file, 'Both', './result/model', scale_offset_record_file, config_json_file)

Parent topic: QAT APIs