python-kerasHow can I view the history of my Python Keras models?

To view the history of your Python Keras models, you can use the model.fit() method. This method takes in the training and validation data, and returns a History object. The History object contains a record of all the metrics that were measured during the training process.

Example code

history = model.fit(X_train, y_train,
                    batch_size=32,
                    epochs=10,
                    validation_data=(X_val, y_val))

The History object has a history attribute that contains a dictionary with all the metrics that were measured during training.

Example code

print(history.history)

Output example

{'val_loss': [0.9031, 0.7456, 0.6854, 0.6458, 0.6185, 0.5989, 0.5839, 0.5722, 0.5646, 0.5584],
 'val_accuracy': [0.7296, 0.7736, 0.7912, 0.8024, 0.8104, 0.816, 0.8208, 0.8236, 0.8264, 0.8296],
 'loss': [1.1817, 0.826, 0.7187, 0.6546, 0.6062, 0.5726, 0.5461, 0.5262, 0.5114, 0.4993],
 'accuracy': [0.6354, 0.7582, 0.7862, 0.8062, 0.822, 0.8324, 0.8412, 0.8468, 0.8518, 0.856]}

You can also access the metrics individually, such as loss and accuracy, using the history.history['loss'] and history.history['accuracy'] methods.

Code explanation

1. model.fit() - method to train the model
2. History object - contains a record of all metrics measured during training
3. history.history - attribute of the History object, contains a dictionary of all metrics measured during training
4. history.history['loss'] and history.history['accuracy'] - methods to access individual metrics

Helpful links

• Keras Documentation - Model Training History

Edit this code on GitHub