ValueError: No data provided for "dense_input"

Question

I am using following simple code to load a csv using tensorflow and perform modeling using keras...

Unable to figure this error!

import tensorflow as tf

train_dataset_fp = tf.keras.utils.get_file(fname=file_path, origin=URL)
columns = ["X","Y"]

features = columns[:-1]
labels = columns[-1]

batch_size = 32

train_dataset = tf.data.experimental.make_csv_dataset(
    train_dataset_fp,
    batch_size,
    column_names = columns,
    label_name= labels,
    num_epochs=1
)

data_iterator = train_dataset.make_one_shot_iterator()

X_train, Y_train = data_iterator.get_next()

from tensorflow import keras

model = keras.Sequential([
    keras.layers.Dense(10, input_shape=[len(X_train)]),
    keras.layers.Dense(1)
])

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

model.summary()

model.fit(X_train, Y_train, epochs=1000, steps_per_epoch=batch_size)

While rest of the code is working fine, i am unable to figure why i am getting dense input error.

The same code works flawlesly if using pandas, i am trying to remove dependencies on other libraries therefore using tensorflow components, but seems to be failing.

_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
dense (Dense)                (None, 10)                30        
_________________________________________________________________
dense_1 (Dense)              (None, 1)                 11        
=================================================================
Total params: 41
Trainable params: 41
Non-trainable params: 0
_________________________________________________________________
Traceback (most recent call last):
  File "/Users/abhinavasrivastava/anaconda3/envs/ml/lib/python3.6/site-packages/tensorflow/python/keras/engine/training_utils.py", line 267, in standardize_input_data
    for x in names
  File "/Users/abhinavasrivastava/anaconda3/envs/ml/lib/python3.6/site-packages/tensorflow/python/keras/engine/training_utils.py", line 267, in <listcomp>
    for x in names
KeyError: 'dense_input'

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "simple_linear_keras.py", line 47, in <module>
    model.fit(X_train, Y_train, epochs=1000, callbacks=[tb], steps_per_epoch=batch_size)
  File "/Users/abhinavasrivastava/anaconda3/envs/ml/lib/python3.6/site-packages/tensorflow/python/keras/engine/training.py", line 1536, in fit
    validation_split=validation_split)
  File "/Users/abhinavasrivastava/anaconda3/envs/ml/lib/python3.6/site-packages/tensorflow/python/keras/engine/training.py", line 992, in _standardize_user_data
    class_weight, batch_size)
  File "/Users/abhinavasrivastava/anaconda3/envs/ml/lib/python3.6/site-packages/tensorflow/python/keras/engine/training.py", line 1117, in _standardize_weights
    exception_prefix='input')
  File "/Users/abhinavasrivastava/anaconda3/envs/ml/lib/python3.6/site-packages/tensorflow/python/keras/engine/training_utils.py", line 271, in standardize_input_data
    'for each key in: ' + str(names))
ValueError: No data provided for "dense_input". Need data for each key in: ['dense_input']

Answer 1

The error No data provided for "dense_input" means that Keras did not get input data at all or not in the expected format, i.e., in the form of an array which in Python means a numpy array.

Supposing everything else works fine it should help to simply add a line to convert X_train and Y_train:

import numpy as np
X_train = np.array(X_train)
Y_train = np.array(Y_train)