CODEWITHSUNDEEP
pythonpython-3.xtensorflowkerasconv-neural-network
Ahmadfromjameedium
Ahmadfromjameedium

Reputation: 35

ValueError: Layer sequential expects 1 inputs, but it received 211 input tensors in tensorflow 2.0

I have a training dataset like this (the number of items in the main list is 211 and the number of numbers in every array is 185):

[np.array([2, 3, 4, ... 5, 4, 6]) ... np.array([3, 4, 5, ... 3, 4, 5])]

and I use this code to train the model:

def create_model():
model = keras.Sequential([
    keras.layers.Flatten(input_shape=(211, 185), name="Input"), 
    keras.layers.Dense(211, activation='relu', name="Hidden_Layer_1"), 
    keras.layers.Dense(185, activation='relu', name="Hidden_Layer_2"), 
    keras.layers.Dense(1, activation='softmax', name="Output"),
])

model.compile(optimizer='adam',
          loss='sparse_categorical_crossentropy',
          metrics=['accuracy'])

return model

but whenever I fit it like this:

model.fit(x=training_data, y=training_labels, epochs=10, validation_data = [training_data,training_labels])

it returns this error:

ValueError: Layer sequential expects 1 inputs, but it received 211 input tensors.

What could possibly be the problem?

Upvotes: 2

Views: 3622

Answers (3)

Ebin Jose Mathew
Ebin Jose Mathew

Reputation: 51

For me it was a silly mistake , i was taking the input in list rather than numpy.ndarray

1.check the type of data format your X_train is in:

type(X_train)

2.If you get output as list or any other format just convert it into numpy.ndarray

X_train = numpy.array(X_train)

Hope this helps Thank you

Upvotes: 2

Nicolas Gervais
Nicolas Gervais

Reputation: 36604

You don't need to flatten your input. If you have 211 samples of shape (185,), this already represents flattened input.

But your initial error is that you can't pass a list of NumPy arrays as input. It needs to be lists of lists or a NumPy array. Try this:

x = np.stack([i.tolist() for i in x])

Then, you made other mistakes. You can't have an output of 1 neuron with a SoftMax activation. It will just output 1, so use "sigmoid". This is also the wrong loss function. If you have two categories, you should use "binary_crossentropy".

Here is a working example of fixing your mistakes, starting from your invalid input:

import tensorflow as tf
import numpy as np

x = [np.random.randint(0, 10, 185) for i in range(211)]
x = np.stack([i.tolist() for i in x])

y = np.random.randint(0, 2, 211)

model = tf.keras.Sequential([ 
    tf.keras.layers.Dense(21, activation='relu', name="Hidden_Layer_1"), 
    tf.keras.layers.Dense(18, activation='relu', name="Hidden_Layer_2"), 
    tf.keras.layers.Dense(1, activation='sigmoid', name="Output"),
])

model.compile(optimizer='adam',
          loss='binary_crossentropy',
          metrics=['accuracy'])


history = model.fit(x=x, y=y, epochs=10)

Upvotes: 5

Andrey
Andrey

Reputation: 6367

You have two errors:

You can not feed a list of arrays. Convert you input to array:

input = np.asarray(input)

You declared input shape of (211, 185). Keras automatically adds batch dimension. So change the shape to (185,):

keras.layers.Flatten(input_shape=(185,), name="Input"),

Upvotes: 1

Related Questions