converts them into `NoOp`s instead.

*   `tf.data`

    * `tf.data` now has an `autotune_options.initial_parallelism` option to
      control the initial parallelism setting used by autotune before the data
      pipeline has started running. The default is 16. A lower value reduces
      initial memory usage, while a higher value improves startup time.

## Keras

distances from the Lens be equal, so that the Rays which converged from
the Lens towards X, and without Refraction, would there have crossed and
diverged again, may by the Refraction of the second Prism be reduced
into Parallelism and diverge no more. For then those Rays will recompose
a beam of white Light XY. If the refracting Angle of either Prism be the
bigger, that Prism must be so much the nearer to the Lens. You will know