If `indices` contains duplicates, the associated `updates` are accumulated
(summed) into the output tensor.

**WARNING**: For floating-point data types, the output may be nondeterministic.
This is because the order in which the updates are applied is nondeterministic
and when floating-point numbers are added in different orders the resulting
numerical approximation error can be slightly different. However, the output

*   Deterministic Op Functionality:

    *   Fix regression in deterministic selection of deterministic cuDNN
        convolution algorithms, a regression that was introduced in v2.5. Note
        that nondeterministic out-of-memory events while selecting algorithms
        could still lead to nondeterminism, although this is very unlikely. This
        additional, unlikely source will be eliminated in a later version.