not supported when used with `tf.distribute.Strategy` outside of
        built-in training loops like `tf.keras` `compile`/`fit`.
    *   Wraps losses passed to the `compile` API (strings and v1 losses) which
        are not instances of v2 `Loss` class in `LossWrapper` class. => All
        losses will now use `SUM_OVER_BATCH_SIZE` reduction as default.

* 🌐 Add Persian translation for `docs/fa/docs/index.md` and tweak right-to-left CSS. PR [#2395](https://github.com/tiangolo/fastapi/pull/2395) by [@mohsen-mahmoodi](https://github.com/mohsen-mahmoodi).

### Internal

* 🔧 Update Jina sponsorship. PR [#5283](https://github.com/tiangolo/fastapi/pull/5283) by [@tiangolo](https://github.com/tiangolo).

return b, nil } } } // isPrime runs the Miller-Rabin Probabilistic Primality Test from // FIPS 186-5, Appendix B.3.1. // // w must be a random odd integer greater than three in big-endian order. // isPrime might return false positives for adversarially chosen values. // // isPrime is not constant-time. func isPrime(w []byte) bool { mr, err := millerRabinSetup(w) if err != nil { // w is zero, one, or even. return false } // Before Miller-Rabin, rule out most composites with trial divisions. for i := 0;...

return b, nil } } } // isPrime runs the Miller-Rabin Probabilistic Primality Test from // FIPS 186-5, Appendix B.3.1. // // w must be a random odd integer greater than three in big-endian order. // isPrime might return false positives for adversarially chosen values. // // isPrime is not constant-time. func isPrime(w []byte) bool { mr, err := millerRabinSetup(w) if err != nil { // w is zero, one, or even. return false } primes, err := bigmod.NewNat().SetBytes(productOfPrimes, mr.w) // If w is too small...