InvalidArgumentError: Несовместимые формы: [1920] и [32,60] [[{{node metrics_3 / acc / Equal}}]] - PullRequest
Новая
       86

InvalidArgumentError: Несовместимые формы: [1920] и [32,60] [[{{node metrics_3 / acc / Equal}}]]

0 голосов
/ 07 апреля 2019

Я нахожусь в обучении модели tagger последовательности Bi-lstm и получаю следующую ошибку с параметрами этой модели: 

model = Sequential()
model.add(Bidirectional(LSTM(units=512, return_sequences=True,
                             recurrent_dropout=0.2, dropout=0.2),input_shape=(max_len,300,)))
model.add(Bidirectional(LSTM(units=512, return_sequences=True,
                             recurrent_dropout=0.2, dropout=0.2)))
model.add(TimeDistributed(Dense(len(tags2index), activation="softmax")))
model.compile(optimizer="adam", loss="sparse_categorical_crossentropy", metrics=["accuracy"])

Вот краткое описание модели: 

Layer (type)                 Output Shape              Param #   
=================================================================
bidirectional_20 (Bidirectio (None, 60, 1024)          3330048   
_________________________________________________________________
bidirectional_21 (Bidirectio (None, 60, 1024)          6295552   
_________________________________________________________________
time_distributed_6 (TimeDist (None, 60, 9)             9225      
=================================================================
Total params: 9,634,825
Trainable params: 9,634,825
Non-trainable params: 0

Вот входные фигуры: 

print(np.array(X_train).shape)
print(y_train.shape)
(6509, 60, 300)
(6509, 60, 1)

Где 6509 - это число примеров, 60 - длина каждого примера, а 300 - размер измерений вектора слова

Старт поезда со следующим кодом: 

batch_size = 32
history = model.fit(np.array(X_train), y_train, validation_data=(np.array(X_eval), y_eval),
                    batch_size=batch_size, epochs=3, verbose=1)

Получение этой трассировки: 

Train on 6509 samples, validate on 1628 samples
Epoch 1/3
---------------------------------------------------------------------------
InvalidArgumentError                      Traceback (most recent call last)
<ipython-input-237-3415d7ea0889> in <module>()
      1 history = model.fit(np.array(X_train), y_train, validation_data=(np.array(X_eval), y_eval),
----> 2                     batch_size=batch_size, epochs=3, verbose=1)

/usr/local/lib/python3.6/dist-packages/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, **kwargs)
   1037                                         initial_epoch=initial_epoch,
   1038                                         steps_per_epoch=steps_per_epoch,
-> 1039                                         validation_steps=validation_steps)
   1040 
   1041     def evaluate(self, x=None, y=None,

/usr/local/lib/python3.6/dist-packages/keras/engine/training_arrays.py in fit_loop(model, f, ins, out_labels, batch_size, epochs, verbose, callbacks, val_f, val_ins, shuffle, callback_metrics, initial_epoch, steps_per_epoch, validation_steps)
    197                     ins_batch[i] = ins_batch[i].toarray()
    198 
--> 199                 outs = f(ins_batch)
    200                 outs = to_list(outs)
    201                 for l, o in zip(out_labels, outs):

/usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py in __call__(self, inputs)
   2713                 return self._legacy_call(inputs)
   2714 
-> 2715             return self._call(inputs)
   2716         else:
   2717             if py_any(is_tensor(x) for x in inputs):

/usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py in _call(self, inputs)
   2673             fetched = self._callable_fn(*array_vals, run_metadata=self.run_metadata)
   2674         else:
-> 2675             fetched = self._callable_fn(*array_vals)
   2676         return fetched[:len(self.outputs)]
   2677 

/usr/local/lib/python3.6/dist-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)

/usr/local/lib/python3.6/dist-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: [1920] vs. [32,60]
     [[{{node metrics_3/acc/Equal}}]]

Я вижу, что 1920 = 32 * 60, но почему TF сравнивает эти размеры?

Кстати, я использую Colab для этой задачи.

Самое запутанное, что он работает с batch_size = 1: (

Не могли бы вы помочь, что я испортил? Заранее спасибо!

...