pytorch训练时验证阶段出现cuda out of memory

验证集需要torch.no_grad()方法：

with torch.no_grad():
        for bt_idx, (imgs, labels) in enumerate(val_loader):
            imgs = imgs.cuda()
            labels = labels.cuda()
            val_outputs = model(imgs)
            val_loss += criterion(val_outputs, labels)
            _,val_prediction = val_outputs.max(1)
            val_correct += val_prediction.eq(labels.cuda()).sum().item()
            val_acc = val_correct/(64*(bt_idx+1))

参考：

1. https://blog.csdn.net/ncc1995/article/details/91126066
2. https://blog.csdn.net/xijuezhu8128/article/details/86594478