*   Added `tf.lookup.experimental.MutableHashTable`, which provides a
        generic mutable hash table implementation.
        *   Compared to `tf.lookup.experimental.DenseHashTable` this offers
            lower overall memory usage, and a cleaner API. It does not require
            specifying a `delete_key` and `empty_key` that cannot be inserted
            into the table.

r4 = addMul64(r4, a3, b1) r4 = addMul64(r4, a4, b0) // After the multiplication, we need to reduce (carry) the five coefficients // to obtain a result with limbs that are at most slightly larger than 2⁵¹, // to respect the Element invariant. // // Overall, the reduction works the same as carryPropagate, except with // wider inputs: we take the carry for each coefficient by shifting it right // by 51, and add it to the limb above it. The top carry is multiplied by 19 // according to the reduction identity...

b2) r4 = addMul(r4, a3, b1) r4 = addMul(r4, a4, b0) // After the multiplication, we need to reduce (carry) the five coefficients // to obtain a result with limbs that are at most slightly larger than 2⁵¹, // to respect the Element invariant. // // Overall, the reduction works the same as carryPropagate, except with // wider inputs: we take the carry for each coefficient by shifting it right // by 51, and add it to the limb above it. The top carry is multiplied by 19 // according to the reduction identity...