google-colab-mnist1

# YOUR CODE SHOULD START HERE
import tensorflow as tf
mnist = tf.keras.datasets.mnist

class myCallback(tf.keras.callbacks.Callback):
  def on_epoch_end(self,epoch,logs={}):
    if(logs.get('acc')>0.99):
      print("Reached 99% accuracy so cancelling training!")
      self.model.stop_training = True
  
# YOUR CODE SHOULD END HERE
#import tensorflow as tf
#mnist = tf.keras.datasets.mnist

(x_train, y_train),(x_test, y_test) = mnist.load_data()
# YOUR CODE SHOULD START HERE
x_train , x_test = x_train/255.0 , x_test/255.0
callbacks = myCallback()
# YOUR CODE SHOULD END HERE
model = tf.keras.models.Sequential([

    tf.keras.layers.Flatten(),
    tf.keras.layers.Dense(512, activation = tf.nn.relu),
    tf.keras.layers.Dense(10, activation = tf.nn.softmax)

])

model.compile(optimizer='adam',
              loss='sparse_categorical_crossentropy',
              metrics=['accuracy'])

# YOUR CODE SHOULD START HERE
model.fit(x_train, y_train, epochs = 10 , callbacks = [callbacks])
# YOUR CODE SHOULD END HERE