paddle.nn.BCELoss的使用问题

ValueError: (InvalidArgument) Cannot get attribute (value) by type float, its type is std::__cxx11::basic_string<char, std::char_traits, std::allocator >. (at /paddle/paddle/fluid/framework/attribute.h:164)

代码：

import paddle

input = paddle.to_tensor( [[0.49922279],[0.46373418]], dtype='float32')
label = paddle.to_tensor([[0.],[1.]], dtype='float32')
output = paddle.nn.BCELoss(input, label)
print(output)

报错：ValueError: (InvalidArgument) Cannot get attribute (value) by type float, its type is std::__cxx11::basic_string<char, std::char_traits, std::allocator >. (at /paddle/paddle/fluid/framework/attribute.h:164)

解决办法：

import paddle

input = paddle.to_tensor( [[0.49922279],[0.46373418]], dtype='float32')
label = paddle.to_tensor([[0.],[1.]], dtype='float32')
bce_loss = paddle.nn.BCELoss()
output = bce_loss(input, label)
print(output)  # [0.65537095]