Appendix: Batch Training
Stay organized with collections Save and categorize content based on your preferences.
Very large datasets may not fit in the memory allocated to your process. In the previous steps, we have set up a pipeline where we bring in the entire dataset in to the memory, prepare the data, and pass the working set to the training function. Instead, Keras provides an alternative training function (fit_generator) that pulls the data in batches. This allows us to apply the transformations in the data pipeline to only a small (a multiple of batch_size) part of the data. During our experiments, we used batching (code in GitHub) for datasets such as DBPedia, Amazon reviews, Ag news, and Yelp reviews.
The following code illustrates how to generate data batches and feed them to fit_generator.
def _data_generator(x, y, num_features, batch_size): """Generates batches of vectorized texts for training/validation. # Arguments x: np.matrix, feature matrix. y: np.ndarray, labels. num_features: int, number of features. batch_size: int, number of samples per batch. # Returns Yields feature and label data in batches. """ num_samples = x.shape[0] num_batches = num_samples // batch_size if num_samples % batch_size: num_batches += 1 while 1: for i in range(num_batches): start_idx = i * batch_size end_idx = (i + 1) * batch_size if end_idx > num_samples: end_idx = num_samples x_batch = x[start_idx:end_idx] y_batch = y[start_idx:end_idx] yield x_batch, y_batch # Create training and validation generators. training_generator = _data_generator( x_train, train_labels, num_features, batch_size) validation_generator = _data_generator( x_val, val_labels, num_features, batch_size) # Get number of training steps. This indicated the number of steps it takes # to cover all samples in one epoch. steps_per_epoch = x_train.shape[0] // batch_size if x_train.shape[0] % batch_size: steps_per_epoch += 1 # Get number of validation steps. validation_steps = x_val.shape[0] // batch_size if x_val.shape[0] % batch_size: validation_steps += 1 # Train and validate model. history = model.fit_generator( generator=training_generator, steps_per_epoch=steps_per_epoch, validation_data=validation_generator, validation_steps=validation_steps, callbacks=callbacks, epochs=epochs, verbose=2) # Logs once per epoch.
Except as otherwise noted, the content of this page is licensed under the Creative Commons Attribution 4.0 License, and code samples are licensed under the Apache 2.0 License. For details, see the Google Developers Site Policies. Java is a registered trademark of Oracle and/or its affiliates.
Last updated 2025-08-25 UTC.
[null,null,["Last updated 2025-08-25 UTC."],[[["\u003cp\u003eKeras' \u003ccode\u003efit_generator\u003c/code\u003e function enables training on very large datasets that exceed memory capacity by processing data in batches.\u003c/p\u003e\n"],["\u003cp\u003eBatching applies data transformations to smaller portions of the dataset, improving efficiency for large datasets like DBPedia, Amazon reviews, Ag news, and Yelp reviews.\u003c/p\u003e\n"],["\u003cp\u003eThe provided \u003ccode\u003e_data_generator\u003c/code\u003e function demonstrates how to create batches of data for use with \u003ccode\u003efit_generator\u003c/code\u003e, yielding feature and label data in manageable chunks.\u003c/p\u003e\n"],["\u003cp\u003eWhen training with \u003ccode\u003efit_generator\u003c/code\u003e, \u003ccode\u003esteps_per_epoch\u003c/code\u003e and \u003ccode\u003evalidation_steps\u003c/code\u003e need to be defined to specify the number of batches needed to cover the entire training and validation datasets, respectively, for one epoch.\u003c/p\u003e\n"]]],[],null,["# Appendix: Batch Training\n\nVery large datasets may not fit in the memory allocated to your process. In the\nprevious steps, we have set up a pipeline where we bring in the entire dataset\nin to the memory, prepare the data, and pass the working set to the training\nfunction. Instead, Keras provides an alternative training function\n([fit_generator](https://keras.io/models/sequential/#fit_generator))\nthat pulls the data in batches. This allows us to apply the transformations in\nthe data pipeline to only a small (a multiple of `batch_size`) part of the data.\nDuring our experiments, we used batching (code in GitHub) for datasets such as\n*DBPedia* , *Amazon reviews* , *Ag news* , and *Yelp reviews*.\n\nThe following code illustrates how to generate data batches and feed them to\n[fit_generator](https://keras.io/models/sequential/#fit_generator). \n\n```scdoc\ndef _data_generator(x, y, num_features, batch_size):\n \"\"\"Generates batches of vectorized texts for training/validation.\n\n # Arguments\n x: np.matrix, feature matrix.\n y: np.ndarray, labels.\n num_features: int, number of features.\n batch_size: int, number of samples per batch.\n\n # Returns\n Yields feature and label data in batches.\n \"\"\"\n num_samples = x.shape[0]\n num_batches = num_samples // batch_size\n if num_samples % batch_size:\n num_batches += 1\n\n while 1:\n for i in range(num_batches):\n start_idx = i * batch_size\n end_idx = (i + 1) * batch_size\n if end_idx \u003e num_samples:\n end_idx = num_samples\n x_batch = x[start_idx:end_idx]\n y_batch = y[start_idx:end_idx]\n yield x_batch, y_batch\n\n# Create training and validation generators.\ntraining_generator = _data_generator(\n x_train, train_labels, num_features, batch_size)\nvalidation_generator = _data_generator(\n x_val, val_labels, num_features, batch_size)\n\n# Get number of training steps. This indicated the number of steps it takes\n# to cover all samples in one epoch.\nsteps_per_epoch = x_train.shape[0] // batch_size\nif x_train.shape[0] % batch_size:\n steps_per_epoch += 1\n\n# Get number of validation steps.\nvalidation_steps = x_val.shape[0] // batch_size\nif x_val.shape[0] % batch_size:\n validation_steps += 1\n\n# Train and validate model.\nhistory = model.fit_generator(\n generator=training_generator,\n steps_per_epoch=steps_per_epoch,\n validation_data=validation_generator,\n validation_steps=validation_steps,\n callbacks=callbacks,\n epochs=epochs,\n verbose=2) # Logs once per epoch.\n```"]]