在另一个函数中调用一个函数，但一个变量是从另一个函数声明/实例化/初始化/分配的

Question

问题

def a(...):
 model = b(...)

我正在运行 a(...) 但未定义模型。

b(...) 看起来像：

def b(...):
 ... 
 model=...
 ...
return model

我的问题：我在 python 中的问题是什么？所以我可以解决它。诸如全局/局部、嵌套函数、递归、静态、在函数内部调用函数，或从另一个函数声明/实例化/初始化/分配之类的东西？

下面是同样的问题，但我的真实代码是因为我用谷歌搜索过，所以我可能需要针对具体案例的帮助。

我运行的是什么：

start_parameter_searching(lrList, momentumList, wdList )

功能：

def start_parameter_searching(lrList, wdList, momentumList):
for i in lrList:
  for k in momentumList:
    for j in wdListt:
      set_train_validation_function(i, k, j)
      trainFunction()

lrList = [0.001, 0.01, 0.1]
wdList = [0.001, 0.01, 0.1]
momentumList = [0.001, 0.01, 0.1]

错误

NameError                                 Traceback (most recent call last)
<ipython-input-20-1d7a642788ca> in <module>()
----> 1 start_parameter_searching(lrList, momentumList, wdList)

1 frames
<ipython-input-17-cd25561c1705> in trainFunction()
     10   for epoch in range(num_epochs):
     11       # train for one epoch, printing every 10 iterations
---> 12       _, loss = train_one_epoch(model, optimizer, data_loader, device, epoch, print_freq=10)
     13       # update the learning rate
     14       lr_scheduler.step()

NameError: name 'model' is not defined

问题

我在def start_parameter_searching(lrList, wdList, momentumList): 中运行def set_train_validation_function(i, k, j):

在def set_train_validation_function(i, k, j): 里面我有model = get_instance_segmentation_model(num_classes) 并且模型没有定义。 get_instance_segmentation_model(num_classes) 可能不会再次被调用/声明/实例化。该函数也在另一个函数中。

所有东西都放在一个伪代码文件中

def set_train_validation_function(i, k, j):
    device = torch.device('cuda') if torch.cuda.is_available() else torch.device('cpu')

# our dataset has two classes only - background and person
num_classes = 2

# get the model using our helper function
model = get_instance_segmentation_model(num_classes)
# move model to the right device
model.to(device)

# construct an optimizer
params = [p for p in model.parameters() if p.requires_grad]
optimizer = torch.optim.SGD(params, lr=i,
                            momentum=k, weight_decay=j)

# and a learning rate scheduler which decreases the learning rate by
# 10x every 3 epochs
lr_scheduler = torch.optim.lr_scheduler.StepLR(optimizer,
                                            step_size=3,
                                            gamma=0.1)


def start_parameter_searching(lrList, wdList, momentumList):
    for i in lrList:
      for k in momentumList:
        for j in wdListt:
          set_train_validation_function(i, k, j)
          trainFunction()

lrList = [0.001, 0.01, 0.1]
wdList = [0.001, 0.01, 0.1]
momentumList = [0.001, 0.01, 0.1]

#start training
start_parameter_searching(lrList, momentumList, wdList )

还有model = get_instance_segmentation_model(num_classes)的问题

def get_instance_segmentation_model(num_classes):
# load an instance segmentation model pre-trained on COCO
model = torchvision.models.detection.maskrcnn_resnet50_fpn(pretrained=True)

# get the number of input features for the classifier
in_features = model.roi_heads.box_predictor.cls_score.in_features
# replace the pre-trained head with a new one
model.roi_heads.box_predictor = FastRCNNPredictor(in_features, num_classes)

# now get the number of input features for the mask classifier
in_features_mask = model.roi_heads.mask_predictor.conv5_mask.in_channels
hidden_layer = 256
# and replace the mask predictor with a new one
model.roi_heads.mask_predictor = MaskRCNNPredictor(in_features_mask,
                                                   hidden_layer,
                                                   num_classes)

return model

Answer 1

听起来您没有返回 model 并继续传递它。

你的意思是：

model = set_train_validation_function(i, k, j)
trainFunction(model)

这意味着def set_train_validation_function(...): 需要return model，然后你需要def trainFunction(model):