返回信息流刚开始学，跑一个模型，结果报如下错误：
```
RuntimeError: element 0 of tensors does not require grad and does not have a grad_fn
```
定位在train函数里，下面是train函数
```
def train(trainloader, model, criterion,optimizer,epoch):
batch_time=AverageMeter()
data_time = AverageMeter()
losses = AverageMeter()
top1 = AverageMeter()
model.train()
end=time.time()
with torch.no_grad():
for i, (input,target) in enumerate(trainloader):
#measure data loading time
data_time.update(time.time()-end)
input,target=input.cuda(),target.cuda()
#compute output
output =model(input)
loss=criterion(output,target)
#measure accuracy and record loss
prec = accuracy(output, target)[0]
losses.update(loss.item(), input.size(0))
top1.update(prec.item(), input.size(0))
#compute gradient and do SGD step
optimizer.zero_grad()
loss.backward()
optimizer.step()
#measure elapsed time
batch_time.update(time.time() - end)
end = time.time()
if i%20==0:
print('Epoch: [{0}][{1}/{2}]\t'
'Time {batch_time.val:.3f} ({batch_time.avg:.3f})\t'
'Data {data_time.val:.3f} ({data_time.avg:.3f})\t'
'Loss {loss.val:.4f} ({loss.avg:.4f})\t'
'Prec {top1.val:.3f}% ({top1.avg:.3f}%)'.format(
epoch, i, len(trainloader), batch_time=batch_time,
data_time=data_time, loss=losses, top1=top1))
del output
```
主程序：
```
def main():
use_gpu=torch.cuda.is_available()
gpu_ids = [0]
device_ids=range(torch.cuda.device_count())
# print(len(device_ids))
if use_gpu:
model=densenet_BC_cifar(depth=100,k=40,num_classes=100)
if not os.path.exists('result'):
os.makedirs('result')
fdir='result/denseNet/cifar100'
if not os.path.exists(fdir):
os.makedirs(fdir)
device = torch.device("cuda:" + str(device_ids[0]))
model=model.to(device) #这里将模型复制到gpu ,默认是cuda('0')
criterion=nn.CrossEntropyLoss().cuda()
optimizer = optim.SGD(model.parameters(), 0.1, momentum=0.9, weight_decay=1e-4)
cudnn.benchmark=True
else:
print('Cuda is not available')
# validate(testloader,model,criterion)
for epoch in range(0,20):
adjust_learning_rate(optimizer,epoch)
#train the epoch
train(trainloader,model,criterion,optimizer,epoch)
#evaluate on test
prec=validate(testloader,model,criterion)
#remember the best and save checkpoint
is_best=pre>best_prec
best_prec=max(prec,best_prec)
save_checkpoint({
'epoch':epoch+1,
'state_dict':model.state_dict(),
'best_prec':best_prec,
'optimizer':optimizer.state_dict(),
},is_best,fdir)
```
恳请各位大佬指点啊[ema13]
这是一条镜像帖。来源：北邮人论坛 / ml-dm / #34877同步于 2019/7/28
该镜像源已超过 30 天没有更新，可能在源站已被删除。
ML_DM机器人发帖
pytorch的bug
Caohf
2019/7/28镜像同步3 回复
订阅后，新回复会通过你的通知中心匿名送达。