python - 在张量中分配行会抛出 "None values not supported"

Question

我编写了一个程序来将行从一个张量更新到另一个张量。以下是我想要实现的目标的一个非常基本的想法

with tf.Session() as sess:
    A = tf.Variable(
        [[[0.2, 0.8, 0.1], [0.0, 1.0, 3.0], [0.0, 1.0, 3.0]], [[0.0, 1.0, 1.0], [0.0, 1.0, 0.5], [0.0, 1.0, 3.0]],
         [[0.0, 1.0, 0.6], [0.0, 1.0, 0.4], [0.0, 1.0, 3.0]]])
    B = tf.Variable(
        [[[1.2, 1.8, 1.1], [1.1, 1.1, 3.1], [1.1, 1.1, 3.1]], [[1.1, 1.1, 1.1], [1.1 ,1.1, 1.5], [1.1, 1.1, 3.1]],
         [[1.0, 1.0, 1.6], [1.1, 1.1, 1.4], [1.1, 1.1, 3.1]]])

    sess.run(tf.global_variables_initializer())
    parent1 = 0
    parent2 = 1

    print("\n A")
    print(sess.run(A[parent2][0]))
    # A=[0. 1. 1.]
    print("\n B")
    print(sess.run(B[parent1][0]))
    # B=[0.2 0.8 0.1]
    print("\n Result")
    B = B[0,0].assign(A[1,0])
    print(sess.run(B[0]))
    #  Result
    # [[0.  1.  1. ]
    #  [1.1 1.1 3.1]
    #  [1.1 1.1 3.1]]

这有效。当我添加第二个运算符时，我的问题出现了:

print("\n Result")
B = B[0,0].assign(A[1,0])
B = B[1,0].assign(A[0,0])

然后我遇到了(底部的完整堆栈跟踪):

ValueError: None values not supported.

似乎由于某种原因，它尝试对张量进行内部转换，并且由于某种原因，传递了 None 类型。我尝试在尝试继续之前评估变量，但这也不起作用这个想法是达到可以使用循环的程度，如下所示:

for i in indexes:
    B = B[parent1][i].assign(A[parent2][i])
    B = B[parent2][i].assign(A[parent1][i])

完整的堆栈跟踪:

Traceback (most recent call last):
  File "tensorflow\python\framework\op_def_library.py", line 527, in _apply_op_helper
    preferred_dtype=default_dtype)
  File "tensorflow\python\framework\ops.py", line 1224, in internal_convert_to_tensor
    ret = conversion_func(value, dtype=dtype, name=name, as_ref=as_ref)
  File "tensorflow\python\framework\constant_op.py", line 305, in _constant_tensor_conversion_function
    return constant(v, dtype=dtype, name=name)
  File "tensorflow\python\framework\constant_op.py", line 246, in constant
    allow_broadcast=True)
  File "tensorflow\python\framework\constant_op.py", line 284, in _constant_impl
    allow_broadcast=allow_broadcast))
  File "tensorflow\python\framework\tensor_util.py", line 454, in make_tensor_proto
    raise ValueError("None values not supported.")
ValueError: None values not supported.

Answer 1

尽管我还没有弄清楚这个错误，但当前解决这个问题的一种方法是使用tf.assign。

op1 = tf.assign(B[0,0], A[1,0])
op2 = tf.assign(B[1,0], A[0,0])

with tf.Session() as sess:
    sess.run(tf.global_variables_initializer())
    sess.run([op1, op2])
    print(B.eval())
#[[[0.  1.  1. ]
#   [1.1 1.1 3.1]
#   [1.1 1.1 3.1]]

#  [[0.2 0.8 0.1]
#   [1.1 1.1 1.5]
#   [1.1 1.1 3.1]]

#  [[1.  1.  1.6]
#   [1.1 1.1 1.4]
#   [1.1 1.1 3.1]]]