Training a neural network on MNIST with Keras

This simple example demonstrates how to plug TensorFlow Datasets (TFDS) into a Keras model.

View on TensorFlow.org Run in Google Colab View source on GitHub Download notebook 
import tensorflow as tf
import tensorflow_datasets as tfds

2025-10-04 11:16:16.699239: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:467] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered
WARNING: All log messages before absl::InitializeLog() is called are written to STDERR
E0000 00:00:1759576576.724018   11705 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered
E0000 00:00:1759576576.732199   11705 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered
W0000 00:00:1759576576.751752   11705 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once.
W0000 00:00:1759576576.751774   11705 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once.
W0000 00:00:1759576576.751777   11705 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once.
W0000 00:00:1759576576.751780   11705 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once.

Step 1: Create your input pipeline

Start by building an efficient input pipeline using advices from:

Load a dataset

Load the MNIST dataset with the following arguments:

  • shuffle_files=True: The MNIST data is only stored in a single file, but for larger datasets with multiple files on disk, it's good practice to shuffle them when training.
  • as_supervised=True: Returns a tuple (img, label) instead of a dictionary {'image': img, 'label': label}.
(ds_train, ds_test), ds_info = tfds.load(
    'mnist',
    split=['train', 'test'],
    shuffle_files=True,
    as_supervised=True,
    with_info=True,
)

2025-10-04 11:16:21.222353: E external/local_xla/xla/stream_executor/cuda/cuda_platform.cc:51] failed call to cuInit: INTERNAL: CUDA error: Failed call to cuInit: CUDA_ERROR_NO_DEVICE: no CUDA-capable device is detected

Build a training pipeline

Apply the following transformations:

def normalize_img(image, label):
  """Normalizes images: `uint8` -> `float32`."""
  return tf.cast(image, tf.float32) / 255., label

ds_train = ds_train.map(
    normalize_img, num_parallel_calls=tf.data.AUTOTUNE)
ds_train = ds_train.cache()
ds_train = ds_train.shuffle(ds_info.splits['train'].num_examples)
ds_train = ds_train.batch(128)
ds_train = ds_train.prefetch(tf.data.AUTOTUNE)

Build an evaluation pipeline

Your testing pipeline is similar to the training pipeline with small differences:

  • You don't need to call tf.data.Dataset.shuffle.
  • Caching is done after batching because batches can be the same between epochs.
ds_test = ds_test.map(
    normalize_img, num_parallel_calls=tf.data.AUTOTUNE)
ds_test = ds_test.batch(128)
ds_test = ds_test.cache()
ds_test = ds_test.prefetch(tf.data.AUTOTUNE)

Step 2: Create and train the model

Plug the TFDS input pipeline into a simple Keras model, compile the model, and train it.

model = tf.keras.models.Sequential([
  tf.keras.layers.Flatten(input_shape=(28, 28)),
  tf.keras.layers.Dense(128, activation='relu'),
  tf.keras.layers.Dense(10)
])
model.compile(
    optimizer=tf.keras.optimizers.Adam(0.001),
    loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True),
    metrics=[tf.keras.metrics.SparseCategoricalAccuracy()],
)

model.fit(
    ds_train,
    epochs=6,
    validation_data=ds_test,
)

/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/keras/src/layers/reshaping/flatten.py:37: UserWarning: Do not pass an `input_shape`/`input_dim` argument to a layer. When using Sequential models, prefer using an `Input(shape)` object as the first layer in the model instead.
  super().__init__(**kwargs)
Epoch 1/6
469/469 ━━━━━━━━━━━━━━━━━━━━ 4s 4ms/step - loss: 0.6073 - sparse_categorical_accuracy: 0.8348 - val_loss: 0.1852 - val_sparse_categorical_accuracy: 0.9468
Epoch 2/6
469/469 ━━━━━━━━━━━━━━━━━━━━ 1s 3ms/step - loss: 0.1737 - sparse_categorical_accuracy: 0.9514 - val_loss: 0.1307 - val_sparse_categorical_accuracy: 0.9619
Epoch 3/6
469/469 ━━━━━━━━━━━━━━━━━━━━ 1s 3ms/step - loss: 0.1192 - sparse_categorical_accuracy: 0.9661 - val_loss: 0.1063 - val_sparse_categorical_accuracy: 0.9697
Epoch 4/6
469/469 ━━━━━━━━━━━━━━━━━━━━ 1s 3ms/step - loss: 0.0948 - sparse_categorical_accuracy: 0.9727 - val_loss: 0.0951 - val_sparse_categorical_accuracy: 0.9723
Epoch 5/6
469/469 ━━━━━━━━━━━━━━━━━━━━ 1s 3ms/step - loss: 0.0744 - sparse_categorical_accuracy: 0.9786 - val_loss: 0.0892 - val_sparse_categorical_accuracy: 0.9738
Epoch 6/6
469/469 ━━━━━━━━━━━━━━━━━━━━ 1s 3ms/step - loss: 0.0599 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.0794 - val_sparse_categorical_accuracy: 0.9772
<keras.src.callbacks.history.History at 0x7fbc39d63880>