update to checkpoint callback options (save_frequency)

introduced the number of batches ('n_batches') option for the save frequency instead of  'batch_size'. Using 'batch_size' works in this tutorial because the length of the training data is 1000 which coincidentally results in a rounded value of ~32 when it is divided by the 'batch_size'. In cases when the number of samples is not 1000, this will result in the model saving at different epoch frequencies other than after every 5 epochs.

the definition of 'save_freq' (https://www.tensorflow.org/api_docs/python/tf/keras/callbacks/ModelCheckpoint#args) clearly refers to the number of batches ('n_batches' in this context) and not the number of samples in a batch ('batch_size').

Author: gdarkwah

site/en/tutorials/keras/save_and_load.ipynb

@@ -385,12 +385,17 @@
         "\n",
         "batch_size = 32\n",
         "\n",
+        "# calculate the number of batches per epoch\n",
+        "import math\n",
+        "n_batches = len(train_images) / batch_size\n",
+        "n_batches = math.ceil(n_batches)    # round up the number of batches to the nearest whole integer\n",
+        "\n",
         "# Create a callback that saves the model's weights every 5 epochs\n",
         "cp_callback = tf.keras.callbacks.ModelCheckpoint(\n",
         "    filepath=checkpoint_path, \n",
         "    verbose=1, \n",
         "    save_weights_only=True,\n",
-        "    save_freq=5*batch_size)\n",
+        "    save_freq=5*n_batches)\n",
         "\n",
         "# Create a new model instance\n",
         "model = create_model()\n",