安装(推荐第二个方法):
pip install thop
pip install --upgrade git+https://github.com/Lyken17/pytorch-OpCounter.git
1.基本用法:
from torchvision.models import resnet50
from thop import profile
model = resnet50()
input = torch.randn(1, 3, 224, 224)
flops, params = profile(model, inputs = (input, ))
2.定义第三方模块的规则:
class YourModule(nn.Module): # your definition
def count_your_model(model, x, y): # your rule here
input = torch.randn(1, 3, 224, 224)
flops, params = profile(model, inputs=(input, ), custom_ops={YourModule:count_your_model})
3.提高输出可读性:
调用thop.clever_format
以提供更好的输出格式
from thop import clever_format
flops, params = clever_format([flops, params], "%.3f")
PyTorch parameters()也可以计算参数量:
from thop import profile
from utils.ShuffleNetV2 import shufflenetv2
import torch
model_shuffle = shufflenetv2(width_mult=0.5)
model = torch.nn.DataParallel(model_shuffle)
total = sum([param.nelement() for param in model.parameters()])
print("Number of parameter: %.2fM" % (total / 1e6))