多阵输入估计器

Question

我正在创建一个带有numpy数组的估计器，用tf.estimator.inputs.numpy_input_fn提供给模型。如下所示：

def input_fun(data):
    x, y = data

    x, y = np.reshape(x, (batch_size, -1, 1)), \
           np.reshape(y, (batch_size, -1, 1))

    return tf.estimator.inputs.numpy_input_fn({'x': x}, y)

def forward(x, params, mode):

    layers = [tf.nn.rnn_cell.LSTMCell(n_neurons) for _ in range(n_layers)]
    cells = tf.nn.rnn_cell.MultiRNNCell(layers)
    outputs, state = tf.nn.dynamic_rnn(cells, x)

    predictions = ...

    return predictions

def model_fn(features, labels, mode, params):
    predict = forward(features, params, mode)

    return tf.estimator.EstimatorSpec(predict , ...)

def experiment_fn(config, params):
    return learn.Experiment(
        estimator = estimator(model_fn,...),
        train_input_fn = lambda: input_fun(train_set),
        eval_input_fn = lambda: input_fun(eval_set))

它会抛出以下内容：

Traceback (most recent call last):

File "", line 1, in runfile('/Experiment.py', wdir='/TensorFlow')

File "C:\Users\hp\Anaconda3\lib\site-packages\spyder\utils\site\sitecustomize.py", line 710, in runfile execfile(filename, namespace)

File "C:\Users\hp\Anaconda3\lib\site-packages\spyder\utils\site\sitecustomize.py", line 101, in execfile exec(compile(f.read(), filename, 'exec'), namespace)

File "/Experiment.py", line 490, in hparams = params

File "C:\Users\hp\Anaconda3\lib\site-packages\tensorflow\contrib\learn\python\learn\learn_runner.py", line 218, in run return _execute_schedule(experiment, schedule)

File "C:\Users\hp\Anaconda3\lib\site-packages\tensorflow\contrib\learn\python\learn\learn_runner.py", line 46, in _execute_schedule return task()

File "C:\Users\hp\Anaconda3\lib\site-packages\tensorflow\contrib\learn\python\learn\experiment.py", line 367, in train hooks=self._train_monitors + extra_hooks)

File "C:\Users\hp\Anaconda3\lib\site-packages\tensorflow\contrib\learn\python\learn\experiment.py", line 807, in _call_train hooks=hooks)

File "C:\Users\hp\Anaconda3\lib\site-packages\tensorflow\python\estimator\estimator.py", line 302, in train loss = self._train_model(input_fn, hooks, saving_listeners)

File "C:\Users\hp\Anaconda3\lib\site-packages\tensorflow\python\estimator\estimator.py", line 711, in _train_model features, labels, model_fn_lib.ModeKeys.TRAIN, self.config)

File "C:\Users\hp\Anaconda3\lib\site-packages\tensorflow\python\estimator\estimator.py", line 694, in _call_model_fn model_fn_results = self._model_fn(features=features, **kwargs)

File "/Experiment.py", line 350, in model_fn predict = forward(features, params, mode)

File "/Experiment.py", line 335, in forward dtype = tf.float32

File "C:\Users\hp\Anaconda3\lib\site-packages\tensorflow\python\ops\rnn.py", line 562, in dynamic_rnn flat_input = [ops.convert_to_tensor(input_) for input_ in flat_input]

File "C:\Users\hp\Anaconda3\lib\site-packages\tensorflow\python\ops\rnn.py", line 562, in flat_input = [ops.convert_to_tensor(input_) for input_ in flat_input]

File "C:\Users\hp\Anaconda3\lib\site-packages\tensorflow\python\framework\ops.py", line 836, in convert_to_tensor as_ref=False)

File "C:\Users\hp\Anaconda3\lib\site-packages\tensorflow\python\framework\ops.py", line 926, in internal_convert_to_tensor ret = conversion_func(value, dtype=dtype, name=name, as_ref=as_ref)

File "C:\Users\hp\Anaconda3\lib\site-packages\tensorflow\python\framework\constant_op.py", line 229, in _constant_tensor_conversion_function return constant(v, dtype=dtype, name=name)

File "C:\Users\hp\Anaconda3\lib\site-packages\tensorflow\python\framework\constant_op.py", line 208, in constant value, dtype=dtype, shape=shape, verify_shape=verify_shape))

File "C:\Users\hp\Anaconda3\lib\site-packages\tensorflow\python\framework\tensor_util.py", line 472, in make_tensor_proto "supported type." % (type(values), values))

TypeError: Failed to convert object of type <class 'function'> to Tensor. Contents: <function numpy_input_fn.<locals>.input_fn at 0x000001AB2B1DBEA0>. Consider casting elements to a supported type.

有人知道为什么吗？在

Answer 1

您应该将单元格列表传递到^{}：

Args:

cells: list of RNNCells that will be composed in this order.

state_is_tuple: If True, accepted and returned states are n-tuples, where n = len(cells). If False, the states are all concatenated along the column axis. This latter behavior will soon be deprecated.

如果您真的想创建一个单层RNN，请将代码更改为

cells = tf.nn.rnn_cell.MultiRNNCell([layers])

。。。或者创建更多层。在

Answer 2

我也有类似的问题。在我的例子中，出现了一个例外，因为在我的模型中（我猜是“forward”，在你的例子中）x被用作张量，但它实际上是一个函数（特别是tf.估计器.输入.numpy_输入_fn）。 我加上这个就知道了：

print(x)
print(type(x))

上面印着这样的东西：

^{pr2}$

我仍然不确定解决问题的正确方法是什么，但我通过做类似的事情来解决它：

^{3}$

希望有帮助