tf.keras.backend.cast_to_floatx

tf.keras.backend.cast_to_floatx(x)

Defined in tensorflow/python/keras/_impl/keras/backend.py.

Cast a Numpy array to the default Keras float type.

Arguments:

  • x: Numpy array.

Returns:

The same Numpy array, cast to its new type.

Example: python >>> from keras import backend as K >>> K.floatx() 'float32' >>> arr = numpy.array([1.0, 2.0], dtype='float64') >>> arr.dtype dtype('float64') >>> new_arr = K.cast_to_floatx(arr) >>> new_arr array([ 1., 2.], dtype=float32) >>> new_arr.dtype dtype('float32')