我正在使用Tensorflow 2 keras，并且当我尝试通过运行来拟合模型时：
model.fit(x=[example_input_batch, example_target_batch], y=np.array(offset_one_timestep), epochs=10, batch_size=1, use_multiprocessing=True)
很好。但是，我只能使用batch_size的1。任何其他值都将返回错误：
---------------------------------------------------------------------------
InvalidArgumentError Traceback (most recent call last)
<ipython-input-10-8c178df9f9b3> in <module>
16
17 #x is input data and y is the target data
---> 18 model.fit(x=[example_input_batch, example_target_batch], y=np.array(offset_one_timestep), epochs=10, batch_size=2, use_multiprocessing=True)
19 # model.fit(x=[trainQ, trainA], y=np.array(offset_one_timestep), epochs=3, batch_size=1)
~/anaconda3/envs/tf2/lib/python3.7/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_freq, max_queue_size, workers, use_multiprocessing, **kwargs)
641 max_queue_size=max_queue_size,
642 workers=workers,
--> 643 use_multiprocessing=use_multiprocessing)
644
645 def evaluate(self,
~/anaconda3/envs/tf2/lib/python3.7/site-packages/tensorflow/python/keras/engine/training_arrays.py in fit(self, model, x, y, batch_size, epochs, verbose, callbacks, validation_split, validation_data, shuffle, class_weight, sample_weight, initial_epoch, steps_per_epoch, validation_steps, validation_freq, **kwargs)
662 validation_steps=validation_steps,
663 validation_freq=validation_freq,
--> 664 steps_name='steps_per_epoch')
665
666 def evaluate(self,
~/anaconda3/envs/tf2/lib/python3.7/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, validation_freq, mode, validation_in_fit, prepared_feed_values_from_dataset, steps_name, **kwargs)
381
382 # Get outputs.
--> 383 batch_outs = f(ins_batch)
384 if not isinstance(batch_outs, list):
385 batch_outs = [batch_outs]
~/anaconda3/envs/tf2/lib/python3.7/site-packages/tensorflow/python/keras/backend.py in __call__(self, inputs)
3508 value = math_ops.cast(value, tensor.dtype)
3509 converted_inputs.append(value)
-> 3510 outputs = self._graph_fn(*converted_inputs)
3511
3512 # EagerTensor.numpy() will often make a copy to ensure memory safety.
~/anaconda3/envs/tf2/lib/python3.7/site-packages/tensorflow/python/eager/function.py in __call__(self, *args, **kwargs)
570 raise TypeError("Keyword arguments {} unknown. Expected {}.".format(
571 list(kwargs.keys()), list(self._arg_keywords)))
--> 572 return self._call_flat(args)
573
574 def _filtered_call(self, args, kwargs):
~/anaconda3/envs/tf2/lib/python3.7/site-packages/tensorflow/python/eager/function.py in _call_flat(self, args)
669 # Only need to override the gradient in graph mode and when we have outputs.
670 if context.executing_eagerly() or not self.outputs:
--> 671 outputs = self._inference_function.call(ctx, args)
672 else:
673 self._register_gradient()
~/anaconda3/envs/tf2/lib/python3.7/site-packages/tensorflow/python/eager/function.py in call(self, ctx, args)
443 attrs=("executor_type", executor_type,
444 "config_proto", config),
--> 445 ctx=ctx)
446 # Replace empty list with None
447 outputs = outputs or None
~/anaconda3/envs/tf2/lib/python3.7/site-packages/tensorflow/python/eager/execute.py in quick_execute(op_name, num_outputs, inputs, attrs, ctx, name)
65 else:
66 message = e.message
---> 67 six.raise_from(core._status_to_exception(e.code, message), None)
68 except TypeError as e:
69 if any(ops._is_keras_symbolic_tensor(x) for x in inputs):
~/anaconda3/envs/tf2/lib/python3.7/site-packages/six.py in raise_from(value, from_value)
InvalidArgumentError: Incompatible shapes: [2,20] vs. [2,20,20]
[[node loss_3/dense_loss/mul (defined at <ipython-input-10-8c178df9f9b3>:18) ]] [Op:__inference_keras_scratch_graph_7008]
Function call stack:
keras_scratch_graph
有人知道这是为什么吗？即使只是删除批处理大小值，也将返回相同的错误。
我的模型是一个非常基本的RNN，带有编码器和解码器。
编辑： 这是我定义的模型：
encoder_input = tf.keras.layers.Input(shape=(20,))
encoder_embedding = tf.keras.layers.Embedding(input_dim=vocab_size, output_dim = 64)
encoder_embedded = encoder_embedding(encoder_input)
output, state_h, state_c = tf.keras.layers.LSTM(128, return_state=True, name='encoder')(encoder_embedded)
encoder_state = [state_h, state_c]
decoder_input = tf.keras.layers.Input(shape=(None,))
decoder_embedded = tf.keras.layers.Embedding(input_dim=vocab_size, output_dim=128)(decoder_input)
decoder_lstm = tf.keras.layers.LSTM(128, return_sequences=True, return_state=True, name='decoder')
decoder_output, _, _ = decoder_lstm(decoder_embedded, initial_state=encoder_state)
output = tf.keras.layers.Dense(20, activation='softmax')(decoder_output)
model = tf.keras.Model(inputs=[encoder_input, decoder_input], outputs=output)
model.compile(loss='categorical_crossentropy', optimizer='adam')
model.summary()