model.fit AttributeError:“元组”对象没有属性“形状”

Posted

技术标签:

【中文标题】model.fit AttributeError:“元组”对象没有属性“形状”【英文标题】:model.fit AttributeError: 'tuple' object has no attribute 'shape' 【发布时间】:2021-12-31 23:49:27 【问题描述】:

我在训练多输入模型时遇到问题。我使用以下代码构建了它:

def create_covn_layers(input_layer):
    input = layers.Conv2D(32, (3,3), input_shape=get_img_input_shape(True))(input_layer)
    covn01 = layers.Conv2D(32, (3, 3))(input)
    acti01 = layers.Activation('relu')(covn01)
    pool01 = layers.MaxPooling2D((2, 2))(acti01)
    covn02 = layers.Conv2D(64, (3, 3))(pool01)
    acti02 = layers.Activation('relu')(covn02)
    pool02 = layers.MaxPooling2D(2, 2)(acti02)
    covn03 = layers.Conv2D(128, (3, 3))(pool02)
    acti02 = layers.Activation('relu')(covn03)
    pool02 = layers.MaxPooling2D(pool_size=(2,2), padding='same')(acti02)
    covn_base = layers.Dropout(0.2)(pool02)

    return covn_base



#flat = layers.Flatten()(pool03)
model_one_input = layers.Input(shape=get_img_input_shape(True))
model_one = create_covn_layers(model_one_input)

model_two_input = layers.Input(shape=get_img_input_shape(True))
model_two = create_covn_layers(model_two_input)

concat_feature_layer = layers.concatenate([model_one, model_two])
flatten_layer = layers.Flatten()(concat_feature_layer)
fully_connected_dense_big = layers.Dense(256, activation='relu')(flatten_layer)
dropout_one = layers.Dropout(0.3)(fully_connected_dense_big)
fully_connected_dense_small = layers.Dense(128, activation='relu')(dropout_one)
dropout_two = layers.Dropout(0.3)(fully_connected_dense_small)
output = layers.Dense(3, activation='softmax')(dropout_two)

model = Model(
    inputs=[model_one_input, model_two_input],
    outputs=output
)

输入层接受以下形状:

batch_size = 18

def get_img_input_shape(for_model=False):
    if for_model:
        return(299,299,3)
    return (299, 299)

[![图像形状层][1]][1]

模型结构:

https://imgur.com/eNtPnjA

我已经构建了一个自定义生成器,它需要两个带有 flowfromdataframe 的生成器并输出两个输入和一个标签。

train_generator_one = ImageDataGenerator(
rescale = 1./255, 
validation_split=0.2
)

train_generator_two = ImageDataGenerator(
rescale = 1./255, 
validation_split=0.2
)

input_1_train_gen = train_generator_one.flow_from_dataframe(
    balanced_eeg_data,
    batch_size=batch_size, 
    target_size=get_img_input_shape(), 
    shuffle=False,
    color_mode="rgb",
    class_mode="categorical",
    subset="training")

input_2_train_gen = train_generator_two.flow_from_dataframe(
    balanced_ecg_data,
    batch_size=batch_size, 
    target_size=get_img_input_shape(), 
    shuffle=False,
    color_mode="rgb",
    class_mode="categorical",
    subset="training")

input_1_validation_gen = train_generator_one.flow_from_dataframe(
    balanced_eeg_data,
    batch_size=batch_size, 
    target_size=get_img_input_shape(), 
    shuffle=False,
    color_mode="rgb",
    class_mode="categorical",
    subset="validation")


input_2_validation_gen = train_generator_two.flow_from_dataframe(
    balanced_ecg_data,
    batch_size=batch_size, 
    target_size=get_img_input_shape(), 
    shuffle=False,
    color_mode="rgb",
    class_mode="categorical",
    subset="validation")

def create_data_generator(data_gen_one, data_gen_two):

    while(True):
        _gen1, _gen1_l = next(data_gen_one)
        _gen2, _gen2_l = next(data_gen_two)

        yield [_gen1, _gen2], [_gen1_l]

multi_train_generator = create_data_generator(
    input_1_train_gen,
    input_2_train_gen
    )

multi_validation_generator = create_data_generator(
    input_1_validation_gen,
    input_2_validation_gen
    )

当我调用 model.fit 但是它给出了一个属性错误:

history = model.fit(
    multi_train_generator,
    epochs=2,
    steps_per_epoch = input_1_train_gen.samples//batch_size, 
    validation_data=multi_validation_generator, 
    validation_steps = input_1_validation_gen.samples//batch_size,
)

---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
/var/folders/0v/m6wt8rqj7s1dcljdyjrdfxmw0000gn/T/ipykernel_84306/4129641024.py in <module>
----> 1 history = model.fit(
      2     multi_train_generator,
      3     epochs=2,
      4     steps_per_epoch = input_1_train_gen.samples//batch_size,
      5     validation_data=multi_validation_generator,

/usr/local/Caskroom/miniforge/base/envs/speciale_01_01/lib/python3.9/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)
   1181                 _r=1):
   1182               callbacks.on_train_batch_begin(step)
-> 1183               tmp_logs = self.train_function(iterator)
   1184               if data_handler.should_sync:
   1185                 context.async_wait()

/usr/local/Caskroom/miniforge/base/envs/speciale_01_01/lib/python3.9/site-packages/tensorflow/python/eager/def_function.py in __call__(self, *args, **kwds)
    887 
    888       with OptionalXlaContext(self._jit_compile):
--> 889         result = self._call(*args, **kwds)
    890 
    891       new_tracing_count = self.experimental_get_tracing_count()

/usr/local/Caskroom/miniforge/base/envs/speciale_01_01/lib/python3.9/site-packages/tensorflow/python/eager/def_function.py in _call(self, *args, **kwds)
    931       # This is the first call of __call__, so we have to initialize.
    932       initializers = []
--> 933       self._initialize(args, kwds, add_initializers_to=initializers)
    934     finally:
    935       # At this point we know that the initialization is complete (or less

/usr/local/Caskroom/miniforge/base/envs/speciale_01_01/lib/python3.9/site-packages/tensorflow/python/eager/def_function.py in _initialize(self, args, kwds, add_initializers_to)
    761     self._graph_deleter = FunctionDeleter(self._lifted_initializer_graph)
    762     self._concrete_stateful_fn = (
--> 763         self._stateful_fn._get_concrete_function_internal_garbage_collected(  # pylint: disable=protected-access
    764             *args, **kwds))
    765 

/usr/local/Caskroom/miniforge/base/envs/speciale_01_01/lib/python3.9/site-packages/tensorflow/python/eager/function.py in _get_concrete_function_internal_garbage_collected(self, *args, **kwargs)
   3048       args, kwargs = None, None
   3049     with self._lock:
-> 3050       graph_function, _ = self._maybe_define_function(args, kwargs)
   3051     return graph_function
   3052 

/usr/local/Caskroom/miniforge/base/envs/speciale_01_01/lib/python3.9/site-packages/tensorflow/python/eager/function.py in _maybe_define_function(self, args, kwargs)
   3442 
   3443           self._function_cache.missed.add(call_context_key)
-> 3444           graph_function = self._create_graph_function(args, kwargs)
   3445           self._function_cache.primary[cache_key] = graph_function
   3446 

/usr/local/Caskroom/miniforge/base/envs/speciale_01_01/lib/python3.9/site-packages/tensorflow/python/eager/function.py in _create_graph_function(self, args, kwargs, override_flat_arg_shapes)
   3277     arg_names = base_arg_names + missing_arg_names
   3278     graph_function = ConcreteFunction(
-> 3279         func_graph_module.func_graph_from_py_func(
   3280             self._name,
   3281             self._python_function,

/usr/local/Caskroom/miniforge/base/envs/speciale_01_01/lib/python3.9/site-packages/tensorflow/python/framework/func_graph.py in func_graph_from_py_func(name, python_func, args, kwargs, signature, func_graph, autograph, autograph_options, add_control_dependencies, arg_names, op_return_value, collections, capture_by_value, override_flat_arg_shapes)
    997         _, original_func = tf_decorator.unwrap(python_func)
    998 
--> 999       func_outputs = python_func(*func_args, **func_kwargs)
   1000 
   1001       # invariant: `func_outputs` contains only Tensors, CompositeTensors,

/usr/local/Caskroom/miniforge/base/envs/speciale_01_01/lib/python3.9/site-packages/tensorflow/python/eager/def_function.py in wrapped_fn(*args, **kwds)
    670         # the function a weak reference to itself to avoid a reference cycle.
    671         with OptionalXlaContext(compile_with_xla):
--> 672           out = weak_wrapped_fn().__wrapped__(*args, **kwds)
    673         return out
    674 

/usr/local/Caskroom/miniforge/base/envs/speciale_01_01/lib/python3.9/site-packages/tensorflow/python/framework/func_graph.py in wrapper(*args, **kwargs)
    984           except Exception as e:  # pylint:disable=broad-except
    985             if hasattr(e, "ag_error_metadata"):
--> 986               raise e.ag_error_metadata.to_exception(e)
    987             else:
    988               raise

AttributeError: in user code:

    /usr/local/Caskroom/miniforge/base/envs/speciale_01_01/lib/python3.9/site-packages/tensorflow/python/keras/engine/training.py:855 train_function  *
        return step_function(self, iterator)
    /usr/local/Caskroom/miniforge/base/envs/speciale_01_01/lib/python3.9/site-packages/tensorflow/python/keras/engine/training.py:845 step_function  **
        outputs = model.distribute_strategy.run(run_step, args=(data,))
    /usr/local/Caskroom/miniforge/base/envs/speciale_01_01/lib/python3.9/site-packages/tensorflow/python/distribute/distribute_lib.py:1285 run
        return self._extended.call_for_each_replica(fn, args=args, kwargs=kwargs)
    /usr/local/Caskroom/miniforge/base/envs/speciale_01_01/lib/python3.9/site-packages/tensorflow/python/distribute/distribute_lib.py:2833 call_for_each_replica
        return self._call_for_each_replica(fn, args, kwargs)
    /usr/local/Caskroom/miniforge/base/envs/speciale_01_01/lib/python3.9/site-packages/tensorflow/python/distribute/distribute_lib.py:3608 _call_for_each_replica
        return fn(*args, **kwargs)
    /usr/local/Caskroom/miniforge/base/envs/speciale_01_01/lib/python3.9/site-packages/tensorflow/python/keras/engine/training.py:838 run_step  **
        outputs = model.train_step(data)
    /usr/local/Caskroom/miniforge/base/envs/speciale_01_01/lib/python3.9/site-packages/tensorflow/python/keras/engine/training.py:800 train_step
        self.compiled_metrics.update_state(y, y_pred, sample_weight)
    /usr/local/Caskroom/miniforge/base/envs/speciale_01_01/lib/python3.9/site-packages/tensorflow/python/keras/engine/compile_utils.py:439 update_state
        self.build(y_pred, y_true)
    /usr/local/Caskroom/miniforge/base/envs/speciale_01_01/lib/python3.9/site-packages/tensorflow/python/keras/engine/compile_utils.py:361 build
        self._metrics = nest.map_structure_up_to(y_pred, self._get_metric_objects,
    /usr/local/Caskroom/miniforge/base/envs/speciale_01_01/lib/python3.9/site-packages/tensorflow/python/util/nest.py:1374 map_structure_up_to
        return map_structure_with_tuple_paths_up_to(
    /usr/local/Caskroom/miniforge/base/envs/speciale_01_01/lib/python3.9/site-packages/tensorflow/python/util/nest.py:1472 map_structure_with_tuple_paths_up_to
        results = [
    /usr/local/Caskroom/miniforge/base/envs/speciale_01_01/lib/python3.9/site-packages/tensorflow/python/util/nest.py:1473 <listcomp>
        func(*args, **kwargs) for args in zip(flat_path_gen, *flat_value_gen)
    /usr/local/Caskroom/miniforge/base/envs/speciale_01_01/lib/python3.9/site-packages/tensorflow/python/util/nest.py:1376 <lambda>
        lambda _, *values: func(*values),  # Discards the path arg.
    /usr/local/Caskroom/miniforge/base/envs/speciale_01_01/lib/python3.9/site-packages/tensorflow/python/keras/engine/compile_utils.py:485 _get_metric_objects
        return [self._get_metric_object(m, y_t, y_p) for m in metrics]
    /usr/local/Caskroom/miniforge/base/envs/speciale_01_01/lib/python3.9/site-packages/tensorflow/python/keras/engine/compile_utils.py:485 <listcomp>
        return [self._get_metric_object(m, y_t, y_p) for m in metrics]
    /usr/local/Caskroom/miniforge/base/envs/speciale_01_01/lib/python3.9/site-packages/tensorflow/python/keras/engine/compile_utils.py:506 _get_metric_object
        y_t_rank = len(y_t.shape.as_list())

    AttributeError: 'tuple' object has no attribute 'shape'

谁能帮助或指出问题出在哪里?

除了路径之外,数据框是相同的。

更新: 我发现 metrics['acc] 正在解决这个问题......非常烦人...... 但是为什么我失败了我还没有发现。 [1]:https://i.stack.imgur.com/AU6HU.png

【问题讨论】:

【参考方案1】:

因此,对于遇到此问题的其他任何人,我都发现了问题。

OBS:使用顺序模型时不会出现此问题...不知道为什么。

但是,当您像我一样对标签进行热编码时:

并使用带有生成器的多输入模型,例如:

那就不要使用metrics=['acc'] 这不起作用,你会得到属性错误。

请参阅以下内容:https://www.tensorflow.org/api_docs/python/tf/keras/metrics/CategoricalAccuracy

使用 tf.keras.metrics.CategoricalAccuracy 这适用于热编码标签。

【讨论】:

以上是关于model.fit AttributeError:“元组”对象没有属性“形状”的主要内容,如果未能解决你的问题,请参考以下文章

model.fit(X,y) 和 model.fit(train_X, train_y) 有啥区别

model.predict() 和 model.fit() 做啥?

keras.models.Model.fit 中的“时代”是啥?

同时运行 model.fit() 和 TensorBoard?

history=model.fit_generator() 为啥 keras 历史是空的?

保存不同时期的 model.fit 历史