Static Quantization: RuntimeError about backend

quantization
1

Net: Alexnet
Dataset: Imagenet
Quantization: Post-training Static Quantization with ‘fbgemm’.

The net:

import os
import sys
import random

import torch
import torch.nn as nn
import torchvision
import torchvision.datasets as datasets
import torchvision.models as models
import torchvision.transforms as transforms

from utils import load_state_dict_from_url

import collections

######## AlexNet model ########

__all__ = ['AlexNet', 'alexnet']

model_urls = { 'alexnet': 'https://download.pytorch.org/models/alexnet-owt-4df8aa71.pth', }

class AlexNet(nn.Module):

    def __init__(self, num_classes=1000):
        super(AlexNet, self).__init__()

        self.features = nn.Sequential(
                nn.Conv2d(3, 64, kernel_size=11, stride=4, padding=2),
                nn.ReLU(inplace=True),
                nn.MaxPool2d(kernel_size=3, stride=2),
                nn.Conv2d(64, 192, kernel_size=5, padding=2),
                nn.ReLU(inplace=True),
                nn.MaxPool2d(kernel_size=3, stride=2),
                nn.Conv2d(192, 384, kernel_size=3, padding=1),
                nn.ReLU(inplace=True),
                nn.Conv2d(384, 256, kernel_size=3, padding=1),
                nn.ReLU(inplace=True),
                nn.Conv2d(256, 256, kernel_size=3, padding=1),
                nn.ReLU(inplace=True),
                nn.MaxPool2d(kernel_size=3, stride=2),
                )
        self.avgpool = nn.AdaptiveAvgPool2d((6, 6))
        self.classifier = nn.Sequential(
                nn.Dropout(),
                nn.Linear(256 * 6 * 6, 4096),
                nn.ReLU(inplace=True),
                nn.Dropout(),
                nn.Linear(4096, 4096),
                nn.ReLU(inplace=True),
                nn.Linear(4096, num_classes),
                )
        # QuantStub converts tensors from floating point to quantized
        self.quant = torch.quantization.QuantStub()
        # DeQuantStub converts tensors from quantized to floating point
        self.dequant = torch.quantization.DeQuantStub()

    def forward(self, x):
        # manually specify where tensors will be converted from floating
        # point to quantized in the quantized model
        x = self.quant(x)
        x = self.features(x)
        x = self.avgpool(x)
        x = torch.flatten(x, 1)
        x = self.classifier(x)
        # manually specify where tensors will be converted from floating
        # point to quantized in the quantized model
        x = self.quant(x)
        return x


def alexnet(pretrained=False, progress=True, **kwargs):
    r"""AlexNet model architecture from the
    `"One weird trick..." <https://arxiv.org/abs/1404.5997>`_ paper.
    Args:
        pretrained (bool): If True, returns a model pre-trained on ImageNet
        progress (bool): If True, displays a progress bar of the download to stderr
    """
    model = AlexNet(**kwargs)
    if pretrained:
        state_dict = load_state_dict_from_url(model_urls['alexnet'], progress=progress)
        model.load_state_dict(state_dict)
    return model

However, I get the following error:

RuntimeError: Could not run 'aten::quantize_per_tensor' with arguments from the 'QuantizedCPU' backend. 'aten::quantize_per_tensor' is only available for these backends: [CPU, CUDA, BackendSelect, Named, AutogradOther, AutogradCPU, AutogradCUDA, AutogradXLA, AutogradPrivateUse1, AutogradPrivateUse2, AutogradPrivateUse3, Tracer, Autocast, Batched, VmapMode].

How to resolve this?

Thanks.

2

This is because there is duplicated QuantStub placed in the model, I think forward should be:

def forward(self, x):
    # manually specify where tensors will be converted from floating
    # point to quantized in the quantized model
    x = self.quant(x)
    x = self.features(x)
    x = self.avgpool(x)
    x = torch.flatten(x, 1)
    x = self.classifier(x)
    # manually specify where tensors will be converted from floating
    # point to quantized in the quantized model
    x = self.dequant(x)
    return x

If you set qconfig for the full model, then all modules will be quantized, the output of self.classifier will be quantized as well. Why do you call self.quant(x) in the end instead of self.dequant(x) in the end?

1 Like