火炬：Dataloader 中的 num_workers 参数引发错误

如何解决火炬：Dataloader 中的 num_workers 参数引发错误

我想使用 GPU 学习我的模型，但是当我尝试在 PyTorch 的 Dataloader 中指向线程数时出现错误。我愿意：

X_train,X_test,y_train,y_test = model_selection.train_test_split(X_reduced,y,test_size=0.4)
X_test,X_val,y_test,y_val = model_selection.train_test_split(X_test,test_size=0.5)

dtype = config['dtype']

train_dataset = torch.utils.data.TensorDataset(torch.tensor(X_train,dtype=dtype,requires_grad=True),torch.tensor(y_train.values,requires_grad=True)) 
valid_dataset = torch.utils.data.TensorDataset(torch.tensor(X_val,torch.tensor(y_val.values,requires_grad=True)) 
train_dataloader = torch.utils.data.DataLoader(train_dataset,shuffle=True,batch_size=config['batch_size'],num_workers=8)
valid_dataloader = torch.utils.data.DataLoader(valid_dataset,num_workers=8)

我也在使用 torchLightning：

class PostflopNet(pl.LightningModule):
    def __init__(self,n_inputs):
        super().__init__()
        
        self.encoder = torch.nn.Sequential(
        torch.nn.Linear(n_inputs,512),torch.nn.ReLU(),torch.nn.Linear(512,128),torch.nn.Linear(128,32),torch.nn.Linear(32,8),torch.nn.Linear(8,1))
        
    def forward(self,x):
        embedding = self.encoder(x)
        return embedding
    
    def configure_optimizers(self):
        optimizer = torch.optim.Adam(self.parameters(),lr=1.0e-2)
        scheduler = torch.optim.lr_scheduler.StepLR(optimizer,step_size=50,gamma=0.6)
        return [optimizer],[scheduler]
        
    def training_step(self,train_batch,batch_idx):
        inputs,labels = train_batch
        predicted = self(inputs)
        loss = torch.nn.MSELoss().__call__(inputs,labels)
        print('train_loss: ',loss)
        print(inputs.shape)
        return loss

model = PostflopNet(X_train.shape[1])
trainer = pl.Trainer(max_epochs=config['n_train_epoch'],gpus=1)

trainer.fit(model,train_dataloader,valid_dataloader)

但我收到此错误：

File "/home/[e-mail]/.local/lib/python3.6/site-packages/torch/utils/data/_utils/collate.py",line 55,in default_collate
    return torch.stack(batch,out=out)
RuntimeError: stack(): functions with out=... arguments don't support automatic differentiation,but one of the arguments requires grad.

如果我删除 Dataloader 中的 num_worker 参数，它就会成功学习。

有什么问题？