tf.keras.backend.cast_to_floatx

Cast a Numpy array to the default Keras float type.

Aliases:

  • tf.compat.v1.keras.backend.cast_to_floatx
  • tf.compat.v2.keras.backend.cast_to_floatx
  • tf.keras.backend.cast_to_floatx
tf.keras.backend.cast_to_floatx(x)

Defined in python/keras/backend.py.

Arguments:

  • x: Numpy array.

Returns:

The same Numpy array, cast to its new type.

Example:

    >>> 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')