完整报错:Expected all tensors to be on the same device, but found at least two devices, cuda:0 and cpu! (when checking argument for argument index in method wrapper__index_select)
环境:python3,游戏本(带gpu)
1. 将model放置到cuda上:
USE_CUDA = torch.cuda.is_available()
if USE_CUDA:
model = model.cuda()
2.单步调试找到报错的代码:
x = self.embedding(x)
#---------------------------------------
loss = Loss(out,y)
3. 解决:在执行以上报错代码前将变量放置到cuda上:
x=x.to(device=torch.device('cuda' if USE_CUDA else 'cpu'))
#---------------------------------------------------------
out=out.to(device=torch.device('cuda' if USE_CUDA else 'cpu'))
y=y.to(device=torch.device('cuda' if USE_CUDA else 'cpu'))
4. 成功运行,无报错