使用 LSTM/GRU 和 Flatten 会引发尺寸不兼容错误答案

【问题标题】：Usage of LSTM/GRU and Flatten throws dimensional incompatibility error使用 LSTM/GRU 和 Flatten 会引发尺寸不兼容错误
【发布时间】：2020-08-17 16:52:33
【问题描述】：

我想利用我在面向数据科学中找到的有前途的神经网络进行案例研究。

我拥有的数据形状是：

X_train:(1200,18,15)
y_train:(1200,18,1)

这里是 NN，其中包含 GRU、Flatten 和 Dense 等层。

def twds_model(layer1=32, layer2=32, layer3=16, dropout_rate=0.5, optimizer='Adam'
                    , learning_rate=0.001, activation='relu', loss='mse'): 
    
    model = Sequential()
    model.add(Bidirectional(GRU(layer1, return_sequences=True),input_shape=(X_train.shape[1],X_train.shape[2])))
    model.add(AveragePooling1D(2))
    model.add(Conv1D(layer2, 3, activation=activation, padding='same', 
               name='extractor'))
    model.add(Flatten())
    model.add(Dense(layer3,activation=activation))
    model.add(Dropout(dropout_rate))
    model.add(Dense(1))
    model.compile(optimizer=optimizer,loss=loss)
    return model

twds_model=twds_model()
print(twds_model.summary())

_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
bidirectional_4 (Bidirection (None, 18, 64)            9216      
_________________________________________________________________
average_pooling1d_1 (Average (None, 9, 64)             0         
_________________________________________________________________
extractor (Conv1D)           (None, 9, 32)             6176      
_________________________________________________________________
flatten_1 (Flatten)          (None, 288)               0         
_________________________________________________________________
dense_3 (Dense)              (None, 16)                4624      
_________________________________________________________________
dropout_4 (Dropout)          (None, 16)                0         
_________________________________________________________________
dense_4 (Dense)              (None, 1)                 17        
=================================================================
Total params: 20,033
Trainable params: 20,033
Non-trainable params: 0
_________________________________________________________________
None

不幸的是，我陷入了一种矛盾的错误陷阱，输入和输出的形状不匹配。这里是上层情况下的错误。

InvalidArgumentError: Incompatible shapes: [144,1] vs. [144,18,1]
     [[{{node loss_2/dense_4_loss/sub}}]]
     [[{{node loss_2/mul}}]]

Train on 10420 samples, validate on 1697 samples
Epoch 1/8

---------------------------------------------------------------------------
InvalidArgumentError                      Traceback (most recent call last)
<ipython-input-30-3f5256ff03ec> in <module>
----> 1 Test_tdws=twds_model.fit(X_train, y_train, epochs=8, batch_size=144, verbose=2, validation_split=(0.14), shuffle=False) #callbacks=[tensorboard])

~\Anaconda3\envs\Tensorflow\lib\site-packages\tensorflow\python\keras\engine\training.py in fit(self, x, y, batch_size, epochs, verbose, callbacks, validation_split, validation_data, shuffle, class_weight, sample_weight, initial_epoch, steps_per_epoch, validation_steps, max_queue_size, workers, use_multiprocessing, **kwargs)
    878           initial_epoch=initial_epoch,
    879           steps_per_epoch=steps_per_epoch,
--> 880           validation_steps=validation_steps)
    881 
    882   def evaluate(self,

~\Anaconda3\envs\Tensorflow\lib\site-packages\tensorflow\python\keras\engine\training_arrays.py in model_iteration(model, inputs, targets, sample_weights, batch_size, epochs, verbose, callbacks, val_inputs, val_targets, val_sample_weights, shuffle, initial_epoch, steps_per_epoch, validation_steps, mode, validation_in_fit, **kwargs)
    327 
    328         # Get outputs.
--> 329         batch_outs = f(ins_batch)
    330         if not isinstance(batch_outs, list):
    331           batch_outs = [batch_outs]

~\Anaconda3\envs\Tensorflow\lib\site-packages\tensorflow\python\keras\backend.py in __call__(self, inputs)
   3074 
   3075     fetched = self._callable_fn(*array_vals,
-> 3076                                 run_metadata=self.run_metadata)
   3077     self._call_fetch_callbacks(fetched[-len(self._fetches):])
   3078     return nest.pack_sequence_as(self._outputs_structure,

~\Anaconda3\envs\Tensorflow\lib\site-packages\tensorflow\python\client\session.py in __call__(self, *args, **kwargs)
   1437           ret = tf_session.TF_SessionRunCallable(
   1438               self._session._session, self._handle, args, status,
-> 1439               run_metadata_ptr)
   1440         if run_metadata:
   1441           proto_data = tf_session.TF_GetBuffer(run_metadata_ptr)

~\Anaconda3\envs\Tensorflow\lib\site-packages\tensorflow\python\framework\errors_impl.py in __exit__(self, type_arg, value_arg, traceback_arg)
    526             None, None,
    527             compat.as_text(c_api.TF_Message(self.status.status)),
--> 528             c_api.TF_GetCode(self.status.status))
    529     # Delete the underlying status object from memory otherwise it stays alive
    530     # as there is a reference to status from this from the traceback due to

InvalidArgumentError: Incompatible shapes: [144,1] vs. [144,18,1]
     [[{{node loss_2/dense_4_loss/sub}}]]
     [[{{node loss_2/mul}}]]

为了完成，y_train 被重塑为 (1200*18,1) 的预期错误：

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-47-2a6d0761b794> in <module>
----> 1 Test_tdws=twds_model.fit(X_train, y_train_flat, epochs=8, batch_size=144, verbose=2, validation_split=(0.14), shuffle=False) #callbacks=[tensorboard])

~\Anaconda3\envs\Tensorflow\lib\site-packages\tensorflow\python\keras\engine\training.py in fit(self, x, y, batch_size, epochs, verbose, callbacks, validation_split, validation_data, shuffle, class_weight, sample_weight, initial_epoch, steps_per_epoch, validation_steps, max_queue_size, workers, use_multiprocessing, **kwargs)
    774         steps=steps_per_epoch,
    775         validation_split=validation_split,
--> 776         shuffle=shuffle)
    777 
    778     # Prepare validation data.

~\Anaconda3\envs\Tensorflow\lib\site-packages\tensorflow\python\keras\engine\training.py in _standardize_user_data(self, x, y, sample_weight, class_weight, batch_size, check_steps, steps_name, steps, validation_split, shuffle)
   2434       # Check that all arrays have the same length.
   2435       if not self._distribution_strategy:
-> 2436         training_utils.check_array_lengths(x, y, sample_weights)
   2437         if self._is_graph_network and not self.run_eagerly:
   2438           # Additional checks to avoid users mistakenly using improper loss fns.

~\Anaconda3\envs\Tensorflow\lib\site-packages\tensorflow\python\keras\engine\training_utils.py in check_array_lengths(inputs, targets, weights)
    454                      'the same number of samples as target arrays. '
    455                      'Found ' + str(list(set_x)[0]) + ' input samples '
--> 456                      'and ' + str(list(set_y)[0]) + ' target samples.')
    457   if len(set_w) > 1:
    458     raise ValueError('All sample_weight arrays should have '

ValueError: Input arrays should have the same number of samples as target arrays. Found 12117 input samples and 218106 target samples

使用的版本是：

Package                Version
---------------------- --------------------
-                      nsorflow-gpu
-ensorflow-gpu         1.13.1
-rotobuf               3.11.3
-umpy                  1.18.1
absl-py                0.9.0
antlr4-python3-runtime 4.8
asn1crypto             1.3.0
astor                  0.7.1
astropy                3.2.1
astunparse             1.6.3
attrs                  19.3.0
audioread              2.1.8
autopep8               1.5.3
backcall               0.1.0
beautifulsoup4         4.9.0
bezier                 0.8.0
bkcharts               0.2
bleach                 3.1.4
blis                   0.2.4
bokeh                  1.1.0
boto3                  1.9.253
botocore               1.12.253
Bottleneck             1.3.2
cachetools             4.1.0
certifi                2020.4.5.1
cffi                   1.14.0
chardet                3.0.4
click                  6.7
cloudpickle            0.5.3
cmdstanpy              0.4.0
color                  0.1
colorama               0.4.3
colorcet               0.9.1
convertdate            2.2.1
copulas                0.2.5
cryptography           2.8
ctgan                  0.2.1
cycler                 0.10.0
cymem                  2.0.2
Cython                 0.29.17
dash                   0.26.0
dash-core-components   0.27.2
dash-html-components   0.11.0
dash-renderer          0.13.2
dask                   0.18.1
dataclasses            0.6
datashader             0.7.0
datashape              0.5.2
datawig                0.1.10
deap                   1.3.0
decorator              4.4.2
defusedxml             0.6.0
deltapy                0.1.1
dill                   0.2.9
distributed            1.22.1
docutils               0.14
entrypoints            0.3
ephem                  3.7.7.1
et-xmlfile             1.0.1
exrex                  0.10.5
Faker                  4.0.3
fastai                 1.0.60
fastprogress           0.2.2
fbprophet              0.6
fire                   0.3.1
Flask                  1.0.2
Flask-Compress         1.4.0
future                 0.17.1
gast                   0.3.3
geojson                2.4.1
geomet                 0.2.0.post2
google-auth            1.14.0
google-auth-oauthlib   0.4.1
google-pasta           0.2.0
gplearn                0.4.1
graphviz               0.13.2
grpcio                 1.29.0
h5py                   2.10.0
HeapDict               1.0.0
holidays               0.10.2
holoviews              1.12.1
html2text              2018.1.9
hyperas                0.4.1
hyperopt               0.1.2
idna                   2.6
imageio                2.5.0
imbalanced-learn       0.3.3
imblearn               0.0
importlib-metadata     1.5.0
impyute                0.0.8
ipykernel              5.1.4
ipython                7.13.0
ipython-genutils       0.2.0
ipywidgets             7.5.1
itsdangerous           0.24
jdcal                  1.4
jedi                   0.16.0
Jinja2                 2.11.1
jmespath               0.9.5
joblib                 0.13.2
jsonschema             3.2.0
jupyter                1.0.0
jupyter-client         6.1.2
jupyter-console        6.0.0
jupyter-core           4.6.3
Keras                  2.2.5
Keras-Applications     1.0.8
Keras-Preprocessing    1.1.2
keras-rectified-adam   0.17.0
kiwisolver             1.2.0
korean-lunar-calendar  0.2.1
librosa                0.7.2
llvmlite               0.32.1
lml                    0.0.1
locket                 0.2.0
LunarCalendar          0.0.9
Markdown               2.6.11
MarkupSafe             1.1.1
matplotlib             3.2.1
missingpy              0.2.0
mistune                0.8.4
mkl-fft                1.0.15
mkl-random             1.1.0
mkl-service            2.3.0
mock                   4.0.2
msgpack                0.5.6
multipledispatch       0.6.0
murmurhash             1.0.2
mxnet                  1.4.1
nb-conda               2.2.1
nb-conda-kernels       2.2.3
nbconvert              5.6.1
nbformat               5.0.4
nbstripout             0.3.7
networkx               2.1
notebook               6.0.3
numba                  0.49.1
numexpr                2.7.1
numpy                  1.19.0
oauthlib               3.1.0
olefile                0.46
opencv-python          4.2.0.34
openpyxl               2.5.5
opt-einsum             3.2.1
packaging              20.3
pandas                 1.0.3
pandasvault            0.0.3
pandocfilters          1.4.2
param                  1.9.0
parso                  0.6.2
partd                  0.3.8
patsy                  0.5.1
pbr                    5.1.3
pickleshare            0.7.5
Pillow                 7.0.0
pip                    20.0.2
plac                   0.9.6
plotly                 4.7.1
plotly-express         0.4.1
preshed                2.0.1
prometheus-client      0.7.1
prompt-toolkit         3.0.4
protobuf               3.11.3
psutil                 5.4.7
py                     1.8.0
pyasn1                 0.4.8
pyasn1-modules         0.2.8
pycodestyle            2.6.0
pycparser              2.20
pyct                   0.4.5
pyensae                1.3.839
pyexcel                0.5.8
pyexcel-io             0.5.7
Pygments               2.6.1
pykalman               0.9.5
PyMeeus                0.3.7
pymongo                3.8.0
pyOpenSSL              19.1.0
pyparsing              2.4.7
pypi                   2.1
pyquickhelper          1.9.3418
pyrsistent             0.16.0
PySocks                1.7.1
pystan                 2.19.1.1
python-dateutil        2.8.1
pytz                   2019.3
pyviz-comms            0.7.2
PyWavelets             0.5.2
pywin32                227
pywinpty               0.5.7
PyYAML                 5.3.1
pyzmq                  18.1.1
qtconsole              4.4.4
rdt                    0.2.1
RegscorePy             1.1
requests               2.23.0
requests-oauthlib      1.3.0
resampy                0.2.2
retrying               1.3.3
rsa                    4.0
s3transfer             0.2.1
scikit-image           0.15.0
scikit-learn           0.23.2
scipy                  1.4.1
sdv                    0.3.2
seaborn                0.9.0
seasonal               0.3.1
Send2Trash             1.5.0
sentinelsat            0.12.2
setuptools             46.3.0
setuptools-git         1.2
six                    1.14.0
sklearn                0.0
sortedcontainers       2.0.4
SoundFile              0.10.3.post1
soupsieve              2.0
spacy                  2.1.8
srsly                  0.1.0
statsmodels            0.9.0
stopit                 1.1.2
sugartensor            1.0.0.2
ta                     0.5.25
tb-nightly             1.14.0a20190603
tblib                  1.3.2
tensorboard            1.13.1
tensorboard-plugin-wit 1.6.0.post3
tensorflow-estimator   1.13.0
tensorflow-gpu         1.13.1
termcolor              1.1.0
terminado              0.8.3
testpath               0.4.4
text-unidecode         1.3
texttable              1.4.0
tf-estimator-nightly   1.14.0.dev2019060501
Theano                 1.0.4
thinc                  7.0.8
threadpoolctl          2.1.0
toml                   0.10.1
toolz                  0.10.0
torch                  1.4.0
torchvision            0.5.0
tornado                6.0.4
TPOT                   0.10.2
tqdm                   4.45.0
traitlets              4.3.3
transforms3d           0.3.1
tsaug                  0.2.1
typeguard              2.7.1
typing                 3.6.6
update-checker         0.16
urllib3                1.22
utm                    0.4.2
wasabi                 0.2.2
wcwidth                0.1.9
webencodings           0.5.1
Werkzeug               1.0.1
wheel                  0.34.2
widgetsnbextension     3.5.1
win-inet-pton          1.1.0
wincertstore           0.2
wrapt                  1.11.2
xarray                 0.10.8
xlrd                   1.1.0
yahoo-historical       0.3.2
zict                   0.1.3
zipp                   2.2.0

非常感谢所有指向正在运行的代码的提示 ;-)！

编辑编辑编辑

更新 tensorflow 和 keras 到最新版本后，我收到以下错误。错误仍然存在，尽管 tensorlfow、CUDA 10.1 和 cudnn 8.0.2 被完全删除并重新安装。该错误是由我的原始代码和 Fallen Aparts 示例代码产生的。

UnknownError:    Fail to find the dnn implementation.
     [[{{node CudnnRNN}}]]
     [[sequential/bidirectional/forward_gru/PartitionedCall]] [Op:__inference_train_function_5731]

Function call stack:
train_function -> train_function -> train_function

None
Epoch 1/4

---------------------------------------------------------------------------
UnknownError                              Traceback (most recent call last)
<ipython-input-1-64eb8afffe02> in <module>
     27     print(twds_model.summary())
     28 
---> 29     twds_model.fit(X_train, y_train, epochs=4)

~\Anaconda3\envs\Tensorflow\lib\site-packages\tensorflow\python\keras\engine\training.py in _method_wrapper(self, *args, **kwargs)
    106   def _method_wrapper(self, *args, **kwargs):
    107     if not self._in_multi_worker_mode():  # pylint: disable=protected-access
--> 108       return method(self, *args, **kwargs)
    109 
    110     # Running inside `run_distribute_coordinator` already.

~\Anaconda3\envs\Tensorflow\lib\site-packages\tensorflow\python\keras\engine\training.py in fit(self, x, y, batch_size, epochs, verbose, callbacks, validation_split, validation_data, shuffle, class_weight, sample_weight, initial_epoch, steps_per_epoch, validation_steps, validation_batch_size, validation_freq, max_queue_size, workers, use_multiprocessing)
   1096                 batch_size=batch_size):
   1097               callbacks.on_train_batch_begin(step)
-> 1098               tmp_logs = train_function(iterator)
   1099               if data_handler.should_sync:
   1100                 context.async_wait()

~\Anaconda3\envs\Tensorflow\lib\site-packages\tensorflow\python\eager\def_function.py in __call__(self, *args, **kwds)
    778       else:
    779         compiler = "nonXla"
--> 780         result = self._call(*args, **kwds)
    781 
    782       new_tracing_count = self._get_tracing_count()

~\Anaconda3\envs\Tensorflow\lib\site-packages\tensorflow\python\eager\def_function.py in _call(self, *args, **kwds)
    838         # Lifting succeeded, so variables are initialized and we can run the
    839         # stateless function.
--> 840         return self._stateless_fn(*args, **kwds)
    841     else:
    842       canon_args, canon_kwds = \

~\Anaconda3\envs\Tensorflow\lib\site-packages\tensorflow\python\eager\function.py in __call__(self, *args, **kwargs)
   2827     with self._lock:
   2828       graph_function, args, kwargs = self._maybe_define_function(args, kwargs)
-> 2829     return graph_function._filtered_call(args, kwargs)  # pylint: disable=protected-access
   2830 
   2831   @property

~\Anaconda3\envs\Tensorflow\lib\site-packages\tensorflow\python\eager\function.py in _filtered_call(self, args, kwargs, cancellation_manager)
   1846                            resource_variable_ops.BaseResourceVariable))],
   1847         captured_inputs=self.captured_inputs,
-> 1848         cancellation_manager=cancellation_manager)
   1849 
   1850   def _call_flat(self, args, captured_inputs, cancellation_manager=None):

~\Anaconda3\envs\Tensorflow\lib\site-packages\tensorflow\python\eager\function.py in _call_flat(self, args, captured_inputs, cancellation_manager)
   1922       # No tape is watching; skip to running the function.
   1923       return self._build_call_outputs(self._inference_function.call(
-> 1924           ctx, args, cancellation_manager=cancellation_manager))
   1925     forward_backward = self._select_forward_and_backward_functions(
   1926         args,

~\Anaconda3\envs\Tensorflow\lib\site-packages\tensorflow\python\eager\function.py in call(self, ctx, args, cancellation_manager)
    548               inputs=args,
    549               attrs=attrs,
--> 550               ctx=ctx)
    551         else:
    552           outputs = execute.execute_with_cancellation(

~\Anaconda3\envs\Tensorflow\lib\site-packages\tensorflow\python\eager\execute.py in quick_execute(op_name, num_outputs, inputs, attrs, ctx, name)
     58     ctx.ensure_initialized()
     59     tensors = pywrap_tfe.TFE_Py_Execute(ctx._handle, device_name, op_name,
---> 60                                         inputs, attrs, num_outputs)
     61   except core._NotOkStatusException as e:
     62     if name is not None:

UnknownError:    Fail to find the dnn implementation.
     [[{{node CudnnRNN}}]]
     [[sequential/bidirectional/forward_gru/PartitionedCall]] [Op:__inference_train_function_5731]

Function call stack:
train_function -> train_function -> train_function

各自的版本列表：

Package                  Version
------------------------ ---------------
-                        nsorflow-gpu
-ensorflow-gpu           2.3.0
-rotobuf                 3.11.3
absl-py                  0.9.0
antlr4-python3-runtime   4.8
asn1crypto               1.3.0
astor                    0.7.1
astropy                  3.2.1
astunparse               1.6.3
attrs                    19.3.0
audioread                2.1.8
autopep8                 1.5.3
backcall                 0.1.0
beautifulsoup4           4.9.0
bezier                   0.8.0
bkcharts                 0.2
bleach                   3.1.4
blis                     0.2.4
bokeh                    1.1.0
boto3                    1.9.253
botocore                 1.12.253
Bottleneck               1.3.2
cachetools               4.1.0
certifi                  2020.4.5.1
cffi                     1.14.0
chardet                  3.0.4
click                    6.7
cloudpickle              0.5.3
cmdstanpy                0.4.0
color                    0.1
colorama                 0.4.3
colorcet                 0.9.1
convertdate              2.2.1
copulas                  0.2.5
cryptography             2.8
ctgan                    0.2.1
cycler                   0.10.0
cymem                    2.0.2
Cython                   0.29.17
dash                     0.26.0
dash-core-components     0.27.2
dash-html-components     0.11.0
dash-renderer            0.13.2
dask                     0.18.1
dataclasses              0.6
datashader               0.7.0
datashape                0.5.2
datawig                  0.1.10
deap                     1.3.0
decorator                4.4.2
defusedxml               0.6.0
deltapy                  0.1.1
dill                     0.2.9
distributed              1.22.1
docutils                 0.14
entrypoints              0.3
ephem                    3.7.7.1
et-xmlfile               1.0.1
exrex                    0.10.5
Faker                    4.0.3
fastai                   1.0.60
fastprogress             0.2.2
fbprophet                0.6
fire                     0.3.1
Flask                    1.0.2
Flask-Compress           1.4.0
future                   0.17.1
gast                     0.3.3
geojson                  2.4.1
geomet                   0.2.0.post2
google-auth              1.14.0
google-auth-oauthlib     0.4.1
google-pasta             0.2.0
gplearn                  0.4.1
graphviz                 0.13.2
grpcio                   1.29.0
h5py                     2.10.0
HeapDict                 1.0.0
holidays                 0.10.2
holoviews                1.12.1
html2text                2018.1.9
hyperas                  0.4.1
hyperopt                 0.1.2
idna                     2.6
imageio                  2.5.0
imbalanced-learn         0.3.3
imblearn                 0.0
importlib-metadata       1.5.0
impyute                  0.0.8
ipykernel                5.1.4
ipython                  7.13.0
ipython-genutils         0.2.0
ipywidgets               7.5.1
itsdangerous             0.24
jdcal                    1.4
jedi                     0.16.0
Jinja2                   2.11.1
jmespath                 0.9.5
joblib                   0.13.2
jsonschema               3.2.0
jupyter                  1.0.0
jupyter-client           6.1.2
jupyter-console          6.0.0
jupyter-core             4.6.3
Keras                    2.4.3
Keras-Applications       1.0.8
Keras-Preprocessing      1.1.2
keras-rectified-adam     0.17.0
kiwisolver               1.2.0
korean-lunar-calendar    0.2.1
librosa                  0.7.2
llvmlite                 0.32.1
lml                      0.0.1
locket                   0.2.0
LunarCalendar            0.0.9
Markdown                 2.6.11
MarkupSafe               1.1.1
matplotlib               3.2.1
missingpy                0.2.0
mistune                  0.8.4
mkl-fft                  1.0.15
mkl-random               1.1.0
mkl-service              2.3.0
mock                     4.0.2
msgpack                  0.5.6
multipledispatch         0.6.0
murmurhash               1.0.2
mxnet                    1.4.1
nb-conda                 2.2.1
nb-conda-kernels         2.2.3
nbconvert                5.6.1
nbformat                 5.0.4
nbstripout               0.3.7
networkx                 2.1
notebook                 6.0.3
numba                    0.49.1
numexpr                  2.7.1
numpy                    1.18.5
oauthlib                 3.1.0
olefile                  0.46
opencv-python            4.2.0.34
openpyxl                 2.5.5
opt-einsum               3.2.1
packaging                20.3
pandas                   1.0.3
pandasvault              0.0.3
pandocfilters            1.4.2
param                    1.9.0
parso                    0.6.2
partd                    0.3.8
patsy                    0.5.1
pbr                      5.1.3
pickleshare              0.7.5
Pillow                   7.0.0
pip                      20.2.2
plac                     0.9.6
plotly                   4.7.1
plotly-express           0.4.1
preshed                  2.0.1
prometheus-client        0.7.1
prompt-toolkit           3.0.4
protobuf                 3.11.3
psutil                   5.4.7
py                       1.8.0
pyasn1                   0.4.8
pyasn1-modules           0.2.8
pycodestyle              2.6.0
pycparser                2.20
pyct                     0.4.5
pyensae                  1.3.839
pyexcel                  0.5.8
pyexcel-io               0.5.7
Pygments                 2.6.1
pykalman                 0.9.5
PyMeeus                  0.3.7
pymongo                  3.8.0
pyOpenSSL                19.1.0
pyparsing                2.4.7
pypi                     2.1
pyquickhelper            1.9.3418
pyrsistent               0.16.0
PySocks                  1.7.1
pystan                   2.19.1.1
python-dateutil          2.8.1
pytz                     2019.3
pyviz-comms              0.7.2
PyWavelets               0.5.2
pywin32                  227
pywinpty                 0.5.7
PyYAML                   5.3.1
pyzmq                    18.1.1
qtconsole                4.4.4
rdt                      0.2.1
RegscorePy               1.1
requests                 2.23.0
requests-oauthlib        1.3.0
resampy                  0.2.2
retrying                 1.3.3
rsa                      4.0
s3transfer               0.2.1
scikit-image             0.15.0
scikit-learn             0.23.2
scipy                    1.4.1
sdv                      0.3.2
seaborn                  0.9.0
seasonal                 0.3.1
Send2Trash               1.5.0
sentinelsat              0.12.2
setuptools               46.3.0
setuptools-git           1.2
six                      1.14.0
sklearn                  0.0
sortedcontainers         2.0.4
SoundFile                0.10.3.post1
soupsieve                2.0
spacy                    2.1.8
srsly                    0.1.0
statsmodels              0.9.0
stopit                   1.1.2
sugartensor              1.0.0.2
ta                       0.5.25
tb-nightly               1.14.0a20190603
tblib                    1.3.2
tensorboard              2.3.0
tensorboard-plugin-wit   1.7.0
tensorflow-gpu           2.3.0
tensorflow-gpu-estimator 2.3.0
termcolor                1.1.0
terminado                0.8.3
testpath                 0.4.4
text-unidecode           1.3
texttable                1.4.0
Theano                   1.0.4
thinc                    7.0.8
threadpoolctl            2.1.0
toml                     0.10.1
toolz                    0.10.0
torch                    1.4.0
torchvision              0.5.0
tornado                  6.0.4
TPOT                     0.10.2
tqdm                     4.45.0
traitlets                4.3.3
transforms3d             0.3.1
tsaug                    0.2.1
typeguard                2.7.1
typing                   3.6.6
update-checker           0.16
urllib3                  1.22
utm                      0.4.2
wasabi                   0.2.2
wcwidth                  0.1.9
webencodings             0.5.1
Werkzeug                 1.0.1
wheel                    0.34.2
widgetsnbextension       3.5.1
win-inet-pton            1.1.0
wincertstore             0.2
wrapt                    1.11.2
xarray                   0.10.8
xlrd                     1.1.0
yahoo-historical         0.3.2
zict                     0.1.3
zipp                     2.2.0

【问题讨论】：

你的数据是什么形状的？我用X_train.shape = (1200, 18, 15) 和y_train.shape = (1200, 18, 1) 和twds_model.fit(X_train, y_train) 进行了检查，效果很好。
@Fallen Apart：它们正是这些形状。你有什么版本的 keras、tensorflow 等？也许这就是问题所在？
我有 2.2.0，但我想版本应该不是问题。确保再一次拥有这些形状。
形状正确（如下）。一定有一些版本差异。但是，在 2.3.0 中也会引发相同的错误。

标签： tensorflow machine-learning multidimensional-array keras lstm

【解决方案1】：

我无法重现您的错误，请检查以下代码是否适合您：

from tensorflow.keras import Sequential
from tensorflow.keras.layers import Conv1D, GRU, Bidirectional, AveragePooling1D, Dense, Flatten, Dropout
import numpy as np


def twds_model(layer1=32, layer2=32, layer3=16, dropout_rate=0.5, optimizer='Adam',
               learning_rate=0.001, activation='relu', loss='mse'):
    model = Sequential()
    model.add(Bidirectional(GRU(layer1, return_sequences=True), input_shape=(X_train.shape[1], X_train.shape[2])))
    model.add(AveragePooling1D(2))
    model.add(Conv1D(layer2, 3, activation=activation, padding='same',
                     name='extractor'))
    model.add(Flatten())
    model.add(Dense(layer3, activation=activation))
    model.add(Dropout(dropout_rate))
    model.add(Dense(1))
    model.compile(optimizer=optimizer, loss=loss)
    return model


if __name__ == '__main__':
    X_train = np.random.rand(1200, 18, 15)
    y_train = np.random.rand(1200, 18, 1)

    twds_model = twds_model()
    print(twds_model.summary())

    twds_model.fit(X_train, y_train, epochs=20)

【讨论】：

对不起，迟到的答案：有一点通常的版本更新恐怖，但现在我回到了更高的版本。我将您的代码复制粘贴到新的 Jupyter Notebook 中：不幸的是，错误完全保持不变。 InvalidArgumentError：不兼容的形状：[32,1] vs. [32,18,1] [[{{node loss/dense_1_loss/MeanSquaredError/sub}}]] [[{{node ConstantFoldingCtrl/loss/dense_1_loss/MeanSquaredError/weighted_loss/广播权重/assert_broadcastable/AssertGuard/Switch_0}}]]
@JDS 是的，现在看起来像 cuDNN 问题，我会尝试 cuDNN v7.6.5。

【解决方案2】：

好的，这对我有用：

Tensorflow 2.3.0
Keras 2.4.2
CUDA 10.1
cuDNN 7.6.5

连同从这个github issue检索到的代码sn-p

import tensorflow as tf
import os
os.environ['TF_CPP_MIN_LOG_LEVEL'] = '3'
os.environ["CUDA_DEVICE_ORDER"] = "PCI_BUS_ID"
os.environ["CUDA_VISIBLE_DEVICES"] = '0' # Set to -1 if CPU should be used CPU = -1 , GPU = 0

gpus = tf.config.experimental.list_physical_devices('GPU')
cpus = tf.config.experimental.list_physical_devices('CPU')

if gpus:
    try:
        # Currently, memory growth needs to be the same across GPUs
        for gpu in gpus:
            tf.config.experimental.set_memory_growth(gpu, True)
        logical_gpus = tf.config.experimental.list_logical_devices('GPU')
        print(len(gpus), "Physical GPUs,", len(logical_gpus), "Logical GPUs")
    except RuntimeError as e:
        # Memory growth must be set before GPUs have been initialized
        print(e)
elif cpus:
    try:
        # Currently, memory growth needs to be the same across GPUs
        logical_cpus= tf.config.experimental.list_logical_devices('CPU')
        print(len(cpus), "Physical CPU,", len(logical_cpus), "Logical CPU")
    except RuntimeError as e:
        # Memory growth must be set before GPUs have been initialized
        print(e)

非常感谢@Fallen Apart，他们一直陪在我身边。如果好奇，您可能还想简要了解一下我的follow up question here ;-)。

【讨论】：