Your Web News in One Place

Help Webnuz

Referal links:

Sign up for GreenGeeks web hosting
December 23, 2022 09:22 pm GMT

How to save trained model in tensorflow ?

To save a trained model in TensorFlow, follow these steps:

  1. Create a tf.keras.Model object or subclass of it.
  2. Train the model using the fit() method.
  3. Create a tf.keras.ModelCheckpoint callback object and pass it to the fit() method as an argument.
  4. Set the save_weights_only parameter to True in the ModelCheckpoint callback object. This will save only the weights of the model, not the entire model structure.
  5. Set the filepath parameter in the ModelCheckpoint callback object to the desired file location where you want to save the model.
  6. Run the fit() method to train the model and save it at the specified file location.

Example:

# Create a modelmodel = tf.keras.Sequential([    tf.keras.layers.Dense(64, activation='relu', input_shape=(10,)),    tf.keras.layers.Dense(64, activation='relu'),    tf.keras.layers.Dense(10, activation='softmax')])# Train the modelmodel.compile(optimizer='adam', loss='sparse_categorical_crossentropy', metrics=['accuracy'])# Create a ModelCheckpoint callback to save the model weightscheckpoint = tf.keras.callbacks.ModelCheckpoint(filepath='/tmp/model.h5', save_weights_only=True)# Train the model and save the weightsmodel.fit(x_train, y_train, epochs=10, callbacks=[checkpoint])

You can then load the saved model weights using the load_weights() method of the model object:

# Load the saved model weightsmodel.load_weights('/tmp/model.h5')# Evaluate the model on the test dataloss, accuracy = model.evaluate(x_test, y_test)print(f'Test loss: {loss}, Test accuracy: {accuracy}')

Hope you like it! Till then happy coding....


Original Link: https://dev.to/anurag629/how-to-save-trained-model-in-tensorflow--2eg6

Share this article:    Share on Facebook
View Full Article

Dev To

An online community for sharing and discovering great ideas, having debates, and making friends

More About this Source Visit Dev To