P30_利用GUP训练（二）

1.代码实战：

（1）调用.to()方法即可。

.to(device)

device = torch.device("cpu")

torch.device("cuda")

点击查看代码

#12.定义训练的设备
# device = torch.device("cpu")
device = torch.device("cuda")


#12.1把网络转移到设备上
dyl = dyl.to(device)

#12.2损失函数-device
loss_fn = loss_fn.to(device)

        #12.3imgs和targets到device上去
        imgs = imgs.to(device)
        targets = targets.to(device)

（2）网络模型和损失函数是不需要重新赋值的，只有数据、图片、标注需要重新赋值。

点击查看代码

#12.定义训练的设备
#torch.device("cpu")
torch.device("cuda")


#12.1把网络转移到设备上
dyl.to(device)