40 lines
935 B
Python
40 lines
935 B
Python
# Copyright (C) 2018-2023 Intel Corporation
|
|
# SPDX-License-Identifier: Apache-2.0
|
|
|
|
#! [dataset]
|
|
import nncf
|
|
import torch
|
|
|
|
calibration_loader = torch.utils.data.DataLoader(...)
|
|
|
|
def transform_fn(data_item):
|
|
images, _ = data_item
|
|
return images
|
|
|
|
calibration_dataset = nncf.Dataset(calibration_loader, transform_fn)
|
|
#! [dataset]
|
|
|
|
#! [quantization]
|
|
import torchvision
|
|
model = torchvision.models.resnet50(pretrained=True)
|
|
|
|
quantized_model = nncf.quantize(model, calibration_dataset)
|
|
#! [quantization]
|
|
|
|
#! [inference]
|
|
import openvino as ov
|
|
|
|
input_fp32 = ... # FP32 model input
|
|
|
|
# convert PyTorch model to OpenVINO model
|
|
ov_quantized_model = ov.convert_model(quantized_model, example_input=input_fp32)
|
|
|
|
# compile the model to transform quantized operations to int8
|
|
model_int8 = ov.compile_model(ov_quantized_model)
|
|
|
|
res = model_int8(input_fp32)
|
|
|
|
# save the model
|
|
ov.save_model(ov_quantized_model, "quantized_model.xml")
|
|
#! [inference]
|