QAT 的不同训练策略
QAT 的不同训练策略#
载入库：
import torch
from torch import nn, jit
from torch.ao.quantization import quantize_qat
from torchvision.models.quantization import mobilenet_v2
# 载入自定义模块
from mod import torchq
from torchq.xinet import CV
from torchq.helper import evaluate, print_size_of_model, load_model
def create_model(num_classes=10,
quantize=False,
pretrained=False):
'''定义模型'''
float_model = mobilenet_v2(pretrained=pretrained,
quantize=quantize)
# 匹配 ``num_classes``
float_model.classifier[1] = nn.Linear(float_model.last_channel,
num_classes)
return float_model
def create_float_model(num_classes,
model_path):
model = create_model(quantize=False,
num_classes=num_classes)
model = load_model(model, model_path)
return model
torch.cuda.empty_cache() # 清空 GPU 缓存
print(torch.cuda.memory_summary()) # 打印显存
|===========================================================================|
| PyTorch CUDA memory summary, device ID 0 |
|---------------------------------------------------------------------------|
| CUDA OOMs: 0 | cudaMalloc retries: 0 |
|===========================================================================|
| Metric | Cur Usage | Peak Usage | Tot Alloc | Tot Freed |
|---------------------------------------------------------------------------|
| Allocated memory | 0 B | 0 B | 0 B | 0 B |
| from large pool | 0 B | 0 B | 0 B | 0 B |
| from small pool | 0 B | 0 B | 0 B | 0 B |
|---------------------------------------------------------------------------|
| Active memory | 0 B | 0 B | 0 B | 0 B |
| from large pool | 0 B | 0 B | 0 B | 0 B |
| from small pool | 0 B | 0 B | 0 B | 0 B |
|---------------------------------------------------------------------------|
| GPU reserved memory | 0 B | 0 B | 0 B | 0 B |
| from large pool | 0 B | 0 B | 0 B | 0 B |
| from small pool | 0 B | 0 B | 0 B | 0 B |
|---------------------------------------------------------------------------|
| Non-releasable memory | 0 B | 0 B | 0 B | 0 B |
| from large pool | 0 B | 0 B | 0 B | 0 B |
| from small pool | 0 B | 0 B | 0 B | 0 B |
|---------------------------------------------------------------------------|
| Allocations | 0 | 0 | 0 | 0 |
| from large pool | 0 | 0 | 0 | 0 |
| from small pool | 0 | 0 | 0 | 0 |
|---------------------------------------------------------------------------|
| Active allocs | 0 | 0 | 0 | 0 |
| from large pool | 0 | 0 | 0 | 0 |
| from small pool | 0 | 0 | 0 | 0 |
|---------------------------------------------------------------------------|
| GPU reserved segments | 0 | 0 | 0 | 0 |
| from large pool | 0 | 0 | 0 | 0 |
| from small pool | 0 | 0 | 0 | 0 |
|---------------------------------------------------------------------------|
| Non-releasable allocs | 0 | 0 | 0 | 0 |
| from large pool | 0 | 0 | 0 | 0 |
| from small pool | 0 | 0 | 0 | 0 |
|---------------------------------------------------------------------------|
| Oversize allocations | 0 | 0 | 0 | 0 |
|---------------------------------------------------------------------------|
| Oversize GPU segments | 0 | 0 | 0 | 0 |
|===========================================================================|
# 设置 warnings
import warnings
warnings.filterwarnings(
action='ignore',
category=DeprecationWarning,
module='.*'
)
warnings.filterwarnings(
action='ignore',
module='torch.ao.quantization'
)
def print_info(model, model_type, num_eval, criterion):
'''打印信息'''
print_size_of_model(model)
top1, top5 = evaluate(model, criterion, test_iter)
print(f'\n{model_type}：\n\t'
f'在 {num_eval} 张图片上评估 accuracy 为: {top1.avg:2.5f}')
超参数设置：
saved_model_dir = 'models/'
float_model_file = 'mobilenet_pretrained_float.pth'
# scripted_qat_model_file = 'mobilenet_qat_scripted_quantized.pth'
# 超参数
float_model_path = saved_model_dir + float_model_file
batch_size = 16
num_classes = 10
num_epochs = 50
learning_rate = 5e-5
ylim = [0.8, 1]
加载数据集：
# 为了 cifar10 匹配 ImageNet，需要将其 resize 到 224
train_iter, test_iter = CV.load_data_cifar10(batch_size=batch_size,
resize=224)
num_eval = sum(len(ys) for _, ys in test_iter)
Files already downloaded and verified
Files already downloaded and verified
打印浮点模型信息：
float_model = create_float_model(num_classes, float_model_path)
model_type = '浮点模型'
criterion = nn.CrossEntropyLoss(reduction="none")
print_info(float_model, model_type, num_eval, criterion)
模型大小：9.187789 MB
浮点模型：
在 10000 张图片上评估 accuracy 为: 94.91000
普通策略：
num_epochs = 1
ylim = [0.85, 1]
device = 'cuda:1'
param_group = True
# 量化参数
is_freeze = False
is_quantized_acc = False
need_qconfig = True # 做一些 QAT 的量化配置工作
# 提供位置参数
args = [train_iter,
test_iter,
learning_rate,
num_epochs,
device,
is_freeze,
is_quantized_acc,
need_qconfig,
param_group,
ylim]
qat_model = create_float_model(num_classes, float_model_path)
qat_model.fuse_model(is_qat=True)
quantized_model = quantize_qat(qat_model, CV.train_fine_tuning, args)
loss 0.013, train acc 0.996, test acc 0.950
352.4 examples/sec on cuda:1
The Kernel crashed while executing code in the the current cell or a previous cell. Please review the code in the cell(s) to identify a possible cause of the failure. Click <a href='https://aka.ms/vscodeJupyterKernelCrash'>here</a> for more info. View Jupyter <a href='command:jupyter.viewOutput'>log</a> for further details.
for name, param in quantized_model.features.named_parameters():
print(param.dtype)
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
torch.float32
quantized_model
QuantizableMobileNetV2(
(features): Sequential(
(0): ConvNormActivation(
(0): ConvBnReLU2d(
(0): Conv2d(3, 32, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
(1): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(2): ReLU()
)
(1): Identity()
(2): Identity()
)
(1): QuantizableInvertedResidual(
(conv): Sequential(
(0): ConvNormActivation(
(0): ConvBnReLU2d(
(0): Conv2d(32, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)
(1): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(2): ReLU()
)
(1): Identity()
(2): Identity()
)
(1): ConvBn2d(
(0): Conv2d(32, 16, kernel_size=(1, 1), stride=(1, 1), bias=False)
(1): BatchNorm2d(16, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
)
(2): Identity()
)
(skip_add): FloatFunctional(
(activation_post_process): Identity()
)
)
(2): QuantizableInvertedResidual(
(conv): Sequential(
(0): ConvNormActivation(
(0): ConvBnReLU2d(
(0): Conv2d(16, 96, kernel_size=(1, 1), stride=(1, 1), bias=False)
(1): BatchNorm2d(96, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(2): ReLU()
)
(1): Identity()
(2): Identity()
)
(1): ConvNormActivation(
(0): ConvBnReLU2d(
(0): Conv2d(96, 96, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), groups=96, bias=False)
(1): BatchNorm2d(96, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(2): ReLU()
)
(1): Identity()
(2): Identity()
)
(2): ConvBn2d(
(0): Conv2d(96, 24, kernel_size=(1, 1), stride=(1, 1), bias=False)
(1): BatchNorm2d(24, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
)
(3): Identity()
)
(skip_add): FloatFunctional(
(activation_post_process): Identity()
)
)
(3): QuantizableInvertedResidual(
(conv): Sequential(
(0): ConvNormActivation(
(0): ConvBnReLU2d(
(0): Conv2d(24, 144, kernel_size=(1, 1), stride=(1, 1), bias=False)
(1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(2): ReLU()
)
(1): Identity()
(2): Identity()
)
(1): ConvNormActivation(
(0): ConvBnReLU2d(
(0): Conv2d(144, 144, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=144, bias=False)
(1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(2): ReLU()
)
(1): Identity()
(2): Identity()
)
(2): ConvBn2d(
(0): Conv2d(144, 24, kernel_size=(1, 1), stride=(1, 1), bias=False)
(1): BatchNorm2d(24, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
)
(3): Identity()
)
(skip_add): FloatFunctional(
(activation_post_process): Identity()
)
)
(4): QuantizableInvertedResidual(
(conv): Sequential(
(0): ConvNormActivation(
(0): ConvBnReLU2d(
(0): Conv2d(24, 144, kernel_size=(1, 1), stride=(1, 1), bias=False)
(1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(2): ReLU()
)
(1): Identity()
(2): Identity()
)
(1): ConvNormActivation(
(0): ConvBnReLU2d(
(0): Conv2d(144, 144, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), groups=144, bias=False)
(1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(2): ReLU()
)
(1): Identity()
(2): Identity()
)
(2): ConvBn2d(
(0): Conv2d(144, 32, kernel_size=(1, 1), stride=(1, 1), bias=False)
(1): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
)
(3): Identity()
)
(skip_add): FloatFunctional(
(activation_post_process): Identity()
)
)
(5): QuantizableInvertedResidual(
(conv): Sequential(
(0): ConvNormActivation(
(0): ConvBnReLU2d(
(0): Conv2d(32, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
(1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(2): ReLU()
)
(1): Identity()
(2): Identity()
)
(1): ConvNormActivation(
(0): ConvBnReLU2d(
(0): Conv2d(192, 192, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=192, bias=False)
(1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(2): ReLU()
)
(1): Identity()
(2): Identity()
)
(2): ConvBn2d(
(0): Conv2d(192, 32, kernel_size=(1, 1), stride=(1, 1), bias=False)
(1): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
)
(3): Identity()
)
(skip_add): FloatFunctional(
(activation_post_process): Identity()
)
)
(6): QuantizableInvertedResidual(
(conv): Sequential(
(0): ConvNormActivation(
(0): ConvBnReLU2d(
(0): Conv2d(32, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
(1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(2): ReLU()
)
(1): Identity()
(2): Identity()
)
(1): ConvNormActivation(
(0): ConvBnReLU2d(
(0): Conv2d(192, 192, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=192, bias=False)
(1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(2): ReLU()
)
(1): Identity()
(2): Identity()
)
(2): ConvBn2d(
(0): Conv2d(192, 32, kernel_size=(1, 1), stride=(1, 1), bias=False)
(1): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
)
(3): Identity()
)
(skip_add): FloatFunctional(
(activation_post_process): Identity()
)
)
(7): QuantizableInvertedResidual(
(conv): Sequential(
(0): ConvNormActivation(
(0): ConvBnReLU2d(
(0): Conv2d(32, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
(1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(2): ReLU()
)
(1): Identity()
(2): Identity()
)
(1): ConvNormActivation(
(0): ConvBnReLU2d(
(0): Conv2d(192, 192, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), groups=192, bias=False)
(1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(2): ReLU()
)
(1): Identity()
(2): Identity()
)
(2): ConvBn2d(
(0): Conv2d(192, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
(1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
)
(3): Identity()
)
(skip_add): FloatFunctional(
(activation_post_process): Identity()
)
)
(8): QuantizableInvertedResidual(
(conv): Sequential(
(0): ConvNormActivation(
(0): ConvBnReLU2d(
(0): Conv2d(64, 384, kernel_size=(1, 1), stride=(1, 1), bias=False)
(1): BatchNorm2d(384, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(2): ReLU()
)
(1): Identity()
(2): Identity()
)
(1): ConvNormActivation(
(0): ConvBnReLU2d(
(0): Conv2d(384, 384, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=384, bias=False)
(1): BatchNorm2d(384, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(2): ReLU()
)
(1): Identity()
(2): Identity()
)
(2): ConvBn2d(
(0): Conv2d(384, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
(1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
)
(3): Identity()
)
(skip_add): FloatFunctional(
(activation_post_process): Identity()
)
)
(9): QuantizableInvertedResidual(
(conv): Sequential(
(0): ConvNormActivation(
(0): ConvBnReLU2d(
(0): Conv2d(64, 384, kernel_size=(1, 1), stride=(1, 1), bias=False)
(1): BatchNorm2d(384, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(2): ReLU()
)
(1): Identity()
(2): Identity()
)
(1): ConvNormActivation(
(0): ConvBnReLU2d(
(0): Conv2d(384, 384, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=384, bias=False)
(1): BatchNorm2d(384, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(2): ReLU()
)
(1): Identity()
(2): Identity()
)
(2): ConvBn2d(
(0): Conv2d(384, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
(1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
)
(3): Identity()
)
(skip_add): FloatFunctional(
(activation_post_process): Identity()
)
)
(10): QuantizableInvertedResidual(
(conv): Sequential(
(0): ConvNormActivation(
(0): ConvBnReLU2d(
(0): Conv2d(64, 384, kernel_size=(1, 1), stride=(1, 1), bias=False)
(1): BatchNorm2d(384, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(2): ReLU()
)
(1): Identity()
(2): Identity()
)
(1): ConvNormActivation(
(0): ConvBnReLU2d(
(0): Conv2d(384, 384, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=384, bias=False)
(1): BatchNorm2d(384, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(2): ReLU()
)
(1): Identity()
(2): Identity()
)
(2): ConvBn2d(
(0): Conv2d(384, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
(1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
)
(3): Identity()
)
(skip_add): FloatFunctional(
(activation_post_process): Identity()
)
)
(11): QuantizableInvertedResidual(
(conv): Sequential(
(0): ConvNormActivation(
(0): ConvBnReLU2d(
(0): Conv2d(64, 384, kernel_size=(1, 1), stride=(1, 1), bias=False)
(1): BatchNorm2d(384, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(2): ReLU()
)
(1): Identity()
(2): Identity()
)
(1): ConvNormActivation(
(0): ConvBnReLU2d(
(0): Conv2d(384, 384, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=384, bias=False)
(1): BatchNorm2d(384, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(2): ReLU()
)
(1): Identity()
(2): Identity()
)
(2): ConvBn2d(
(0): Conv2d(384, 96, kernel_size=(1, 1), stride=(1, 1), bias=False)
(1): BatchNorm2d(96, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
)
(3): Identity()
)
(skip_add): FloatFunctional(
(activation_post_process): Identity()
)
)
(12): QuantizableInvertedResidual(
(conv): Sequential(
(0): ConvNormActivation(
(0): ConvBnReLU2d(
(0): Conv2d(96, 576, kernel_size=(1, 1), stride=(1, 1), bias=False)
(1): BatchNorm2d(576, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(2): ReLU()
)
(1): Identity()
(2): Identity()
)
(1): ConvNormActivation(
(0): ConvBnReLU2d(
(0): Conv2d(576, 576, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=576, bias=False)
(1): BatchNorm2d(576, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(2): ReLU()
)
(1): Identity()
(2): Identity()
)
(2): ConvBn2d(
(0): Conv2d(576, 96, kernel_size=(1, 1), stride=(1, 1), bias=False)
(1): BatchNorm2d(96, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
)
(3): Identity()
)
(skip_add): FloatFunctional(
(activation_post_process): Identity()
)
)
(13): QuantizableInvertedResidual(
(conv): Sequential(
(0): ConvNormActivation(
(0): ConvBnReLU2d(
(0): Conv2d(96, 576, kernel_size=(1, 1), stride=(1, 1), bias=False)
(1): BatchNorm2d(576, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(2): ReLU()
)
(1): Identity()
(2): Identity()
)
(1): ConvNormActivation(
(0): ConvBnReLU2d(
(0): Conv2d(576, 576, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=576, bias=False)
(1): BatchNorm2d(576, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(2): ReLU()
)
(1): Identity()
(2): Identity()
)
(2): ConvBn2d(
(0): Conv2d(576, 96, kernel_size=(1, 1), stride=(1, 1), bias=False)
(1): BatchNorm2d(96, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
)
(3): Identity()
)
(skip_add): FloatFunctional(
(activation_post_process): Identity()
)
)
(14): QuantizableInvertedResidual(
(conv): Sequential(
(0): ConvNormActivation(
(0): ConvBnReLU2d(
(0): Conv2d(96, 576, kernel_size=(1, 1), stride=(1, 1), bias=False)
(1): BatchNorm2d(576, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(2): ReLU()
)
(1): Identity()
(2): Identity()
)
(1): ConvNormActivation(
(0): ConvBnReLU2d(
(0): Conv2d(576, 576, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), groups=576, bias=False)
(1): BatchNorm2d(576, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(2): ReLU()
)
(1): Identity()
(2): Identity()
)
(2): ConvBn2d(
(0): Conv2d(576, 160, kernel_size=(1, 1), stride=(1, 1), bias=False)
(1): BatchNorm2d(160, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
)
(3): Identity()
)
(skip_add): FloatFunctional(
(activation_post_process): Identity()
)
)
(15): QuantizableInvertedResidual(
(conv): Sequential(
(0): ConvNormActivation(
(0): ConvBnReLU2d(
(0): Conv2d(160, 960, kernel_size=(1, 1), stride=(1, 1), bias=False)
(1): BatchNorm2d(960, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(2): ReLU()
)
(1): Identity()
(2): Identity()
)
(1): ConvNormActivation(
(0): ConvBnReLU2d(
(0): Conv2d(960, 960, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=960, bias=False)
(1): BatchNorm2d(960, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(2): ReLU()
)
(1): Identity()
(2): Identity()
)
(2): ConvBn2d(
(0): Conv2d(960, 160, kernel_size=(1, 1), stride=(1, 1), bias=False)
(1): BatchNorm2d(160, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
)
(3): Identity()
)
(skip_add): FloatFunctional(
(activation_post_process): Identity()
)
)
(16): QuantizableInvertedResidual(
(conv): Sequential(
(0): ConvNormActivation(
(0): ConvBnReLU2d(
(0): Conv2d(160, 960, kernel_size=(1, 1), stride=(1, 1), bias=False)
(1): BatchNorm2d(960, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(2): ReLU()
)
(1): Identity()
(2): Identity()
)
(1): ConvNormActivation(
(0): ConvBnReLU2d(
(0): Conv2d(960, 960, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=960, bias=False)
(1): BatchNorm2d(960, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(2): ReLU()
)
(1): Identity()
(2): Identity()
)
(2): ConvBn2d(
(0): Conv2d(960, 160, kernel_size=(1, 1), stride=(1, 1), bias=False)
(1): BatchNorm2d(160, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
)
(3): Identity()
)
(skip_add): FloatFunctional(
(activation_post_process): Identity()
)
)
(17): QuantizableInvertedResidual(
(conv): Sequential(
(0): ConvNormActivation(
(0): ConvBnReLU2d(
(0): Conv2d(160, 960, kernel_size=(1, 1), stride=(1, 1), bias=False)
(1): BatchNorm2d(960, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(2): ReLU()
)
(1): Identity()
(2): Identity()
)
(1): ConvNormActivation(
(0): ConvBnReLU2d(
(0): Conv2d(960, 960, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=960, bias=False)
(1): BatchNorm2d(960, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(2): ReLU()
)
(1): Identity()
(2): Identity()
)
(2): ConvBn2d(
(0): Conv2d(960, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)
(1): BatchNorm2d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
)
(3): Identity()
)
(skip_add): FloatFunctional(
(activation_post_process): Identity()
)
)
(18): ConvNormActivation(
(0): ConvBnReLU2d(
(0): Conv2d(320, 1280, kernel_size=(1, 1), stride=(1, 1), bias=False)
(1): BatchNorm2d(1280, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(2): ReLU()
)
(1): Identity()
(2): Identity()
)
)
(classifier): Sequential(
(0): Dropout(p=0.2, inplace=False)
(1): Linear(in_features=1280, out_features=10, bias=True)
)
(quant): QuantStub()
(dequant): DeQuantStub()
)
qmodel_file = 'mobilenet_quantization_quantized.pth'
torch.save(quantized_model.state_dict(), saved_model_dir + qmodel_file)
冻结前几次训练的量化器以及观测器：
args[5] = True
args[6] = False
qat_model = create_float_model(num_classes, float_model_path)
quantized_model = quantize_qat(qat_model, CV.train_fine_tuning, args)
loss 0.004, train acc 0.999, test acc 0.951
237.7 examples/sec on cuda:2
输出量化精度：
args[6] = True
args[5] = False
qat_model = create_float_model(num_classes, float_model_path)
quantized_model = quantize_qat(qat_model, CV.train_fine_tuning, args)
loss 0.004, train acc 0.999, test acc 0.953
239.0 examples/sec on cuda:2
冻结前几次训练的观测器并且生成量化精度：
args[5] = True
args[6] = True
qat_model = create_float_model(num_classes, float_model_path)
quantized_model = quantize_qat(qat_model, CV.train_fine_tuning, args)
loss 0.004, train acc 0.999, test acc 0.951
238.1 examples/sec on cuda:2
