*   XLA:CPU and XLA:GPU devices are no longer registered by default. Use
        `TF_XLA_FLAGS=--tf_xla_enable_xla_devices` if you really need them, but
        this flag will eventually be removed in subsequent releases.

*   `tf.keras`:

    *   The `steps_per_execution` argument in `model.compile()` is no longer
        experimental; if you were passing `experimental_steps_per_execution`,

construction of deterministic nonces. The nonce must be 96 bits, the // first 32 bits must be an encoding of the module name, and the last 64 bits // must be a counter. The starting value of the counter is set on the first call // to Seal, and each subsequent call must increment it as a big-endian uint64. // If the counter reaches the starting value minus one, Seal will panic. // // This complies with FIPS 140-3 IG C.H Scenario 3. func NewGCMWithCounterNon(cipher *aes.Block) (*GCMWithCounterNonce, error)...