ValueError: Tried to convert 'y' to a tensor and failed. Error: None values not supported

Question

不起作用:

from tensorflow.python.keras.layers import Input, Dense
from tensorflow.python.keras.models import Model
from tensorflow.python.keras.optimizers import Nadam
import numpy as np

ipt = Input(shape=(4,))
out = Dense(1, activation='sigmoid')(ipt)

model = Model(ipt, out)
model.compile(optimizer=Nadam(lr=1e-4), loss='binary_crossentropy')

X = np.random.randn(32,4)
Y = np.random.randint(0,2,(32,1))
model.train_on_batch(X,Y)

WORKS：从上面的导入中删除 .python。

这是怎么回事，如何解决？

附加信息：

CUDA 10.0.130，cuDNN 7.4.2，Python3.7.4，Windows10
tensorflow、tensorflow-gpu v2.0.0 和 Keras 2.3.0 通过 pip，所有其他通过 Anaconda 3
根据 DEBUG 1，我注意到 pip 安装 r2.0 分支而不是 master；用 master 手动覆盖本地 tensorflow_core.python 文件夹会破坏一切 - 但对 select 这样做 - 很少有文件不会，但错误仍然存在

调试 1：文件差异

这适用于我的本地安装，而不是 TF 的 Github 分支 master 或 r2.0； Github files 由于某种原因缺少 api/_v2:

from tensorflow import keras
print(keras.__file__)
from tensorflow.python import keras
print(keras.__file__)

[1] D:\Anaconda\envs\tf2_env\lib\site-packages\tensorflow_core\python\keras\api\_v2\keras\__init__.py
[2] D:\Anaconda\envs\tf2_env\lib\site-packages\tensorflow_core\python\keras\__init__.py

查看 __init__ 的每个 Optimizer：

# [1]
from tensorflow.python.keras.optimizer_v2.optimizer_v2 import OptimizerV2 as Optimizer
# [2]
from tensorflow.python.keras import optimizers

# in python.keras.optimizers.py:
# all imports are from tensorflow.python
class Optimizer(object): # <--- does NOT use optimizer_v2 for Optimizer

这似乎是问题的根源，如下所示：

from tensorflow.python.keras.layers import Input, Dense
from tensorflow.python.keras.models import Model
from tensorflow.keras.optimizers import Nadam

然而，这很奇怪，因为直接 import keras 也不使用 optimizer_v2，尽管 keras.optimizers 中 Optimizer 的定义确实不同。

调试 2：执行差异

并行调试，虽然两者使用相同的 training.py，但执行差异相当快：

### TF.KERAS
    if self._experimental_run_tf_function: #  TRUE
### TF.PYTHON.KERAS
    if self._experimental_run_tf_function: #  FALSE

前者在最终失败之前继续调用 training_v2_utils.train_on_batch(...) 和 returns，后者 self._standardize_user_data(...) 和其他人。

调试 3（+ 解决方案？）：失败线

if None in grads: # <-- in traceback

在其正上方插入 print(None in grads) 会产生完全相同的错误 - 因此，它似乎与 TF2 可迭代操作相关 - 这有效：

if any([g is None for g in grads]): # <-- works; similar but not equivalent Python logic

不确定它是否是一个完整的修复程序，仍在调试 -- 更新：启动了 Github Pull Request

完整错误跟踪:

  File "<ipython-input-1-2db039c052cf>", line 20, in <module>
    model.train_on_batch(X,Y)
  File "D:\Anaconda\envs\tf2_env\lib\site-packages\tensorflow_core\python\keras\engine\training.py", line 1017, in train_on_batch
    self._make_train_function()
  File "D:\Anaconda\envs\tf2_env\lib\site-packages\tensorflow_core\python\keras\engine\training.py", line 2116, in _make_train_function
    params=self._collected_trainable_weights, loss=self.total_loss)
  File "D:\Anaconda\envs\tf2_env\lib\site-packages\tensorflow_core\python\keras\optimizers.py", line 653, in get_updates
    grads = self.get_gradients(loss, params)
  File "D:\Anaconda\envs\tf2_env\lib\site-packages\tensorflow_core\python\keras\optimizers.py", line 92, in get_gradients
    if None in grads:
  File "D:\Anaconda\envs\tf2_env\lib\site-packages\tensorflow_core\python\ops\math_ops.py", line 1336, in tensor_equals
    return gen_math_ops.equal(self, other)
  File "D:\Anaconda\envs\tf2_env\lib\site-packages\tensorflow_core\python\ops\gen_math_ops.py", line 3626, in equal
    name=name)
  File "D:\Anaconda\envs\tf2_env\lib\site-packages\tensorflow_core\python\framework\op_def_library.py", line 545, in _apply_op_helper
    (input_name, err))

ValueError: Tried to convert 'y' to a tensor and failed. Error: None values not supported.

Answer 1

这是一个错误，我的 pull request fix was approved (but isn't yet merged). In the meantime, you can make the change manually, as here. Also, tf.python.keras isn't always meant to be used, if 。

更新：拉取请求现在是 merged。

为什么有效：None in grads 与 any(g == None for g in grads) 相同；问题是，g 可能是一个 tf.Tensor/tf.Variable，其中 .__eq__ 定义为仅对张量进行操作，因此必须使用 is None 代替。

from tensorflow.keras.layers import Input, Dense
from tensorflow.keras.models import Model
import numpy as np

ipt = Input((16,))
out = Dense(16)(ipt)
model = Model(ipt, out)
model.compile('adam', 'mse')

x = y = np.random.randn(32, 16)
model.train_on_batch(x, y)

W = model.optimizer.weights
W[0] == None

>>> ValueError: Attempt to convert a value (None) with an unsupported type 
    (<class 'NoneType'>) to a Tensor.

检查源代码：

from inspect import getsource
print(getsource(W[0].__eq__))

def __eq__(self, other):
    """Compares two variables element-wise for equality."""
    if ops.Tensor._USE_EQUALITY and ops.executing_eagerly_outside_functions():
        return gen_math_ops.equal(self, other, incompatible_shape_error=False)
    else:
        # In legacy graph mode, tensor equality is object equality
        return self is other

Answer 2

也许你应该更正你的导入

from tensorflow.keras.layers import Input, Dense
from tensorflow.keras.models import Model
from tensorflow.keras.optimizers import Nadam

ValueError: Tried to convert 'y' to a tensor and failed. Error: None values not supported

ValueError: Tried to convert 'y' to a tensor and failed. Error: None values not supported

python

keras

tensorflow

tensorflow2.0