Tensorflow模型拟合ValueError:图层顺序期望一个输入，但它收到520个输入张量

Question

我试着用Tensorflow训练一个模型。我正在使用tf.data.experimental.make_csv_dataset读取一个巨大的csv文件。

这是我的代码：

导入：

import tensorflow as tf
from tensorflow import keras
from tensorflow.keras import layers
from tensorflow.keras.layers.experimental import preprocessing

LABEL_COLUMN = 'venda_qtde'

将csv读入tf.data.Dataset:

def get_dataset(file_path, **kwargs):
  dataset = tf.data.experimental.make_csv_dataset(
      file_path,
      batch_size=4096, 
      na_value="?",
      label_name=LABEL_COLUMN,
      num_epochs=1,
      ignore_errors=False,
      shuffle=False,
      **kwargs)
  return dataset

构建模型实例：

def build_model():
  model = None
  model = keras.Sequential([
    layers.Dense(520, activation='relu'),
    layers.Dense(520, activation='relu'),
    layers.Dense(520, activation='relu'),
    layers.Dense(1)
  ])

  model.compile(loss='mean_squared_error',
                optimizer='adam',
                metrics=['mae'])
  return model

执行函数的

：

ds_treino = get_dataset('data/processed/curva_a/curva_a_train.csv')
nn_model = build_model()
nn_model.fit(ds_treino, epochs=10)

但是，当调用fit函数时，我会得到错误：

ValueError: in user code:

    /home/machine-learning/.virtualenvs/jupyter-n5c7sT9n/lib/python3.8/site-packages/tensorflow/python/keras/engine/training.py:806 train_function  *
        return step_function(self, iterator)
    /home/machine-learning/.virtualenvs/jupyter-n5c7sT9n/lib/python3.8/site-packages/tensorflow/python/keras/engine/training.py:796 step_function  **
        outputs = model.distribute_strategy.run(run_step, args=(data,))
    /home/machine-learning/.virtualenvs/jupyter-n5c7sT9n/lib/python3.8/site-packages/tensorflow/python/distribute/distribute_lib.py:1211 run
        return self._extended.call_for_each_replica(fn, args=args, kwargs=kwargs)
    /home/machine-learning/.virtualenvs/jupyter-n5c7sT9n/lib/python3.8/site-packages/tensorflow/python/distribute/distribute_lib.py:2585 call_for_each_replica
        return self._call_for_each_replica(fn, args, kwargs)
    /home/machine-learning/.virtualenvs/jupyter-n5c7sT9n/lib/python3.8/site-packages/tensorflow/python/distribute/distribute_lib.py:2945 _call_for_each_replica
        return fn(*args, **kwargs)
    /home/machine-learning/.virtualenvs/jupyter-n5c7sT9n/lib/python3.8/site-packages/tensorflow/python/keras/engine/training.py:789 run_step  **
        outputs = model.train_step(data)
    /home/machine-learning/.virtualenvs/jupyter-n5c7sT9n/lib/python3.8/site-packages/tensorflow/python/keras/engine/training.py:747 train_step
        y_pred = self(x, training=True)
    /home/machine-learning/.virtualenvs/jupyter-n5c7sT9n/lib/python3.8/site-packages/tensorflow/python/keras/engine/base_layer.py:975 __call__
        input_spec.assert_input_compatibility(self.input_spec, inputs,
    /home/machine-learning/.virtualenvs/jupyter-n5c7sT9n/lib/python3.8/site-packages/tensorflow/python/keras/engine/input_spec.py:155 assert_input_compatibility
        raise ValueError('Layer ' + layer_name + ' expects ' +

    ValueError: Layer sequential expects 1 inputs, but it received 520 input tensors. Inputs received: ...

我的数据集有519个特征，一个标签和大约17M行。

有人能帮我做错事吗？

Answer 1

首先，查看tf.data.Dataset返回的张量的形状，然后尝试设置第一个密集层的input_shape，如下所示：

model = keras.Sequential([
    layers.Dense(520, activation='relu', input_shape=(1, 519)),
    layers.Dense(520, activation='relu'),
    layers.Dense(520, activation='relu'),
    layers.Dense(1)
  ])

或者显式地添加输入层。

或设置第一致密层对应于特征数的神经元数目(519)

还可以阅读它们非常棒的文档：https://www.tensorflow.org/api_docs/python/tf/keras/序列