Python 自定义层中的输入形状为“无”

我正在Tensorflow 2.1中构建自己的层，并在自定义模型中使用它。然而，当我试图学习一些东西时，当第一次调用时，层试图构建自己，它需要输入形状来完成。据我所知，它应该计算它，因为它得到了一个实际的输入，但似乎输入大小是零

我的问题是：我做错了什么，如何纠正

下面我附上一个例子来重现这个问题

我的代码（MinimarNcell从tensorflow网站复制）：

我得到的错误是：

Traceback (most recent call last):
  File "/home/.../test.py", line 64, in <module>
    model.fit(x,y,epochs=10, batch_size=1)
  File "/home/.../.venv/lib/python3.6/site-packages/tensorflow_core/python/keras/engine/training.py", line 819, in fit
    use_multiprocessing=use_multiprocessing)
  File "/home/.../.venv/lib/python3.6/site-packages/tensorflow_core/python/keras/engine/training_v2.py", line 235, in fit
    use_multiprocessing=use_multiprocessing)
  File "/home/.../.venv/lib/python3.6/site-packages/tensorflow_core/python/keras/engine/training_v2.py", line 593, in _process_training_inputs
    use_multiprocessing=use_multiprocessing)
  File "/home/.../.venv/lib/python3.6/site-packages/tensorflow_core/python/keras/engine/training_v2.py", line 646, in _process_inputs
    x, y, sample_weight=sample_weights)
  File "/home/.../.venv/lib/python3.6/site-packages/tensorflow_core/python/keras/engine/training.py", line 2346, in _standardize_user_data
    all_inputs, y_input, dict_inputs = self._build_model_with_inputs(x, y)
  File "/home/.../.venv/lib/python3.6/site-packages/tensorflow_core/python/keras/engine/training.py", line 2572, in _build_model_with_inputs
    self._set_inputs(cast_inputs)
  File "/home/.../.venv/lib/python3.6/site-packages/tensorflow_core/python/keras/engine/training.py", line 2659, in _set_inputs
    outputs = self(inputs, **kwargs)
  File "/home/.../.venv/lib/python3.6/site-packages/tensorflow_core/python/keras/engine/base_layer.py", line 773, in __call__
    outputs = call_fn(cast_inputs, *args, **kwargs)
  File "/home/.../.venv/lib/python3.6/site-packages/tensorflow_core/python/autograph/impl/api.py", line 237, in wrapper
    raise e.ag_error_metadata.to_exception(e)
TypeError: in converted code:

    /home/.../test.py:36 call  *
        out=self.minimalrnn(input)
    /home/.../.venv/lib/python3.6/site-packages/tensorflow_core/python/keras/engine/base_layer.py:818 __call__
        self._maybe_build(inputs)
    /home/.../.venv/lib/python3.6/site-packages/tensorflow_core/python/keras/engine/base_layer.py:2116 _maybe_build
        self.build(input_shapes)
    /home/.../test.py:14 build
        self.kernel = self.add_weight(shape=(input_shape[-1], self.units),

    TypeError: 'NoneType' object is not subscriptable

回溯（最近一次呼叫最后一次）：
文件“/home/../test.py”，第64行，在
模型拟合（x，y，历次=10，批次大小=1）
文件“/home/../.venv/lib/python3.6/site packages/tensorflow_core/python/keras/engine/training.py”，第819行，以适合的形式
使用多处理=使用多处理）
文件“/home/...venv/lib/python3.6/site packages/tensorflow_core/python/keras/engine/training_v2.py”，第235行
使用多处理=使用多处理）
文件“/home/...venv/lib/python3.6/site packages/tensorflow\u core/python/keras/engine/training\u v2.py”，第593行，进程中的培训输入
使用多处理=使用多处理）
文件“/home/../.venv/lib/python3.6/site packages/tensorflow\u core/python/keras/engine/training\u v2.py”，第646行，在进程输入中
x、 y，样本重量=样本重量）
文件“/home/...venv/lib/python3.6/site packages/tensorflow\u core/python/keras/engine/training.py”，第2346行，在用户数据中
所有输入，y输入，dict输入=自我。用输入（x，y）建立模型
文件“/home/...venv/lib/python3.6/site packages/tensorflow\u core/python/keras/engine/training.py”，第2572行，在带有输入的构建模型中
自设置输入（转换输入）
文件“/home/../.venv/lib/python3.6/site packages/tensorflow\u core/python/keras/engine/training.py”，第2659行，在集合输入中
输出=自身（输入，**kwargs）
文件“/home/...venv/lib/python3.6/site packages/tensorflow\u core/python/keras/engine/base\u layer.py”，第773行，在调用中__
输出=调用fn（转换输入，*args，**kwargs）
文件“/home/...venv/lib/python3.6/site packages/tensorflow_core/python/autograph/impl/api.py”，第237行，在包装器中
将e.ag\u错误\u元数据引发到\u异常（e）
TypeError:在转换的代码中：
/home/../test.py:36呼叫*
输出=自最小RNN（输入）
/home/...venv/lib/python3.6/site packages/tensorflow\u core/python/keras/engine/base\u layer.py:818\u调用__
自我构建（输入）
/home/...venv/lib/python3.6/site packages/tensorflow\u core/python/keras/engine/base\u layer.py:2116\u maybe\u build
自我构建（输入形状）
/home/../test.py:14构建
self.kernel=self.add_weight（shape=（输入_shape[-1]，self.units），
TypeError:“非类型”对象不可下标
有一个输入错误（
输入
应该是
输入
）。
输入
是一个内置函数（）

out=self.minimalrnn（输入）
应该是
out=self.minimalrnn（输入）
哦，哇。我现在觉得自己真的很笨。谢谢你发现这个打字错误。它没有解决输入大小的问题，但还有另一个问题-但我认为这是单独的问题，我正在链接它以获取信息：
Traceback (most recent call last):
  File "/home/.../test.py", line 64, in <module>
    model.fit(x,y,epochs=10, batch_size=1)
  File "/home/.../.venv/lib/python3.6/site-packages/tensorflow_core/python/keras/engine/training.py", line 819, in fit
    use_multiprocessing=use_multiprocessing)
  File "/home/.../.venv/lib/python3.6/site-packages/tensorflow_core/python/keras/engine/training_v2.py", line 235, in fit
    use_multiprocessing=use_multiprocessing)
  File "/home/.../.venv/lib/python3.6/site-packages/tensorflow_core/python/keras/engine/training_v2.py", line 593, in _process_training_inputs
    use_multiprocessing=use_multiprocessing)
  File "/home/.../.venv/lib/python3.6/site-packages/tensorflow_core/python/keras/engine/training_v2.py", line 646, in _process_inputs
    x, y, sample_weight=sample_weights)
  File "/home/.../.venv/lib/python3.6/site-packages/tensorflow_core/python/keras/engine/training.py", line 2346, in _standardize_user_data
    all_inputs, y_input, dict_inputs = self._build_model_with_inputs(x, y)
  File "/home/.../.venv/lib/python3.6/site-packages/tensorflow_core/python/keras/engine/training.py", line 2572, in _build_model_with_inputs
    self._set_inputs(cast_inputs)
  File "/home/.../.venv/lib/python3.6/site-packages/tensorflow_core/python/keras/engine/training.py", line 2659, in _set_inputs
    outputs = self(inputs, **kwargs)
  File "/home/.../.venv/lib/python3.6/site-packages/tensorflow_core/python/keras/engine/base_layer.py", line 773, in __call__
    outputs = call_fn(cast_inputs, *args, **kwargs)
  File "/home/.../.venv/lib/python3.6/site-packages/tensorflow_core/python/autograph/impl/api.py", line 237, in wrapper
    raise e.ag_error_metadata.to_exception(e)
TypeError: in converted code:

    /home/.../test.py:36 call  *
        out=self.minimalrnn(input)
    /home/.../.venv/lib/python3.6/site-packages/tensorflow_core/python/keras/engine/base_layer.py:818 __call__
        self._maybe_build(inputs)
    /home/.../.venv/lib/python3.6/site-packages/tensorflow_core/python/keras/engine/base_layer.py:2116 _maybe_build
        self.build(input_shapes)
    /home/.../test.py:14 build
        self.kernel = self.add_weight(shape=(input_shape[-1], self.units),

    TypeError: 'NoneType' object is not subscriptable

class RNNXModel(Model):
    def __init__(self, size):
        super(RNNXModel, self).__init__()
        self.minimalrnn=MinimalRNNCell(size)

    def call(self, inputs):
        out=self.minimalrnn(inputs)  # changed from `input`
        return out