Pytorch 错误：输入应为标量类型 Long 但发现 Float

Question

我正在尝试创建一个深度学习算法来玩蛇。我正在尝试使用 PyTorch 来实现这一点。这是我的（混乱，稍后会修复）代码的 sn-p：

## DOUBLE Q DEEP LEARNING NETWORK
class SnakeNet(nn.Module):
    """mini cnn structure
  input -> (conv2d + relu) x 3 -> flatten -> (dense + relu) x 2 -> output
  """
    def __init__(self, input_dim, output_dim):
        super().__init__()

        self.online = nn.Sequential(
            # nn.Conv2d(in_channels=input_dim, out_channels=32, kernel_size=8, stride=4),
            # nn.ReLU(),
            # nn.Conv2d(in_channels=32, out_channels=64, kernel_size=4, stride=2),
            # nn.ReLU(),
            # nn.Conv2d(in_channels=64, out_channels=64, kernel_size=3, stride=1),
            # nn.ReLU(),
            # nn.Flatten(),
            # nn.Linear(3136, 512),
            # nn.ReLU(),
            # nn.Linear(512, output_dim),
            nn.Linear(input_dim, 200),
            nn.Linear(200, 20),
            nn.Linear(20, 50),
            nn.Linear(50, output_dim),
        )

        self.target = copy.deepcopy(self.online)

        # Q_target parameters are frozen.
        for p in self.target.parameters():
            p.requires_grad = False

    def forward(self, input, model):
        input = input.long()
        if model == "online":
            return self.online(input)
        elif model == "target":
            return self.target(input)

# EXPLOIT
        else:
            state = torch.tensor(state)
            state = state.unsqueeze(0)
            action_values = self.net(state, model="online")
            dir = torch.argmax(action_values, axis=1).item()

第 221 行出现错误：action_values = self.net(state, model="online")

声明我的输入（状态）是一个浮点数，尽管它是一个 tensorLong，我通过打印 type() 证明了这一点。在建议添加state = state.type.tensorLong() 之前，这不起作用，主要是因为它已经很久了。

错误：

Traceback (most recent call last):
  File "snakeGame.py", line 324, in <module>
    prev_location, action = snake.act(current_state)
  File "snakeGame.py", line 222, in act
    action_values = self.net(state, model="online")
  File "/Users/gavinhartog/opt/anaconda3/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1102, in _call_impl
    return forward_call(*input, **kwargs)
  File "snakeGame.py", line 57, in forward
    return self.online(input)
  File "/Users/gavinhartog/opt/anaconda3/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1102, in _call_impl
    return forward_call(*input, **kwargs)
  File "/Users/gavinhartog/opt/anaconda3/lib/python3.8/site-packages/torch/nn/modules/container.py", line 141, in forward
    input = module(input)
  File "/Users/gavinhartog/opt/anaconda3/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1102, in _call_impl
    return forward_call(*input, **kwargs)
  File "/Users/gavinhartog/opt/anaconda3/lib/python3.8/site-packages/torch/nn/modules/linear.py", line 103, in forward
    return F.linear(input, self.weight, self.bias)
  File "/Users/gavinhartog/opt/anaconda3/lib/python3.8/site-packages/torch/nn/functional.py", line 1848, in linear
    return torch._C._nn.linear(input, weight, bias)
RuntimeError: expected scalar type Long but found Float

这是在torch.tensor之前的状态的原始内容和形状：

[[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 3, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], 
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 3, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], etc, etc, etc

我尝试过不同的东西，比如 Conv2d 和不同的损失函数，但都是同样的错误。提前致谢。

Answer 1

这个错误有点令人困惑。但我想你已经将state 类型转换为float 而不是long

state = state.float()

因为nn.Linear 总是需要浮点数。

Answer 2

你能用这个再试一次吗：

stats=np.array([1,2,3,4,5,6])
print(type(stats))
stats=torch.tensor(stats).type(torch.LongTensor)
state = stats.unsqueeze(0)
print(type(stats))