How to Fix AttributeError: ‘Sequential’ Object Has No Attribute ‘predict_classes’

Diagram of Sequential Object Has No Attribute predict_classes


AttributeError: ‘Sequential’ Object Has No Attribute ‘predict_classes’ error occurs because “you are using a version of Keras or TensorFlow where ‘predict_classes’ no longer exists.” In Keras, the ‘predict_classes’ was deprecated in Keras 2.3.0 and completely removed in Keras 2.4.0.

Reproducing the error

from keras.models import Sequential

model = Sequential()

y_pred = model.predict_classes(validation_data)


AttributeError: 'Sequential' object has no attribute 'predict_classes'

How to Fix it?

To fix the AttributeError: ‘Sequential’ Object Has No Attribute ‘predict_classes’, you can use the “argmax” function from the numpy library along with the “predict” method from your “Sequential” object. This combination will provide you with the same output as ‘predict_classes’.

import numpy as np
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense

# Define the model
model = Sequential()
model.add(Dense(32, activation='relu', input_dim=100))
model.add(Dense(10, activation='softmax'))

# Compile the model
model.compile(optimizer='adam', loss='categorical_crossentropy',

# Assume 'data' is your input for prediction
data = np.random.random((500, 100))

# Use the 'predict' method to get the class probabilities
probabilities = model.predict(data)

# Use numpy's 'argmax' function to get the class predictions
class_predictions = np.argmax(probabilities, axis=-1)


16/16 [==============================] - 0s 279us/step

After running the above code in my console, you can see that the error is successfully fixed!

Alternatively, you can downgrade your Keras or TensorFlow version.

In Python’s deep learning library, Keras, ‘Sequential’ refers to a linear stack of neural network layers that can be created and manipulated easily.

The best practice for making class predictions in recent versions of Keras and TensorFlow is using the ‘predict’ method and the ‘argmax’ function for the numpy library.

That’s it.

Leave a Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.