See https://arxiv.org/abs/2206.14286 for the algorithm details.
This op is only optimized on TPU currently.
  }];

  let arguments = (ins
    Arg<TensorOf<[TF_Bfloat16, TF_Float16, TF_Float32]>, [{Array to search. Must be at least 1-D of the floating type}]>:$input,

    ConfinedAttr<I64Attr, [IntMinValue<0>]>:$k,
    DefaultValuedOptionalAttr<I64Attr, "-1">:$reduction_dimension,

    1.x behavior.

*   `CUDNN_INSTALL_PATH`, `TENSORRT_INSTALL_PATH`, `NCCL_INSTALL_PATH`,
    `NCCL_HDR_PATH` are deprecated. Use `TF_CUDA_PATHS` instead which supports a
    comma-separated list of base paths that are searched to find CUDA libraries
    and headers.

Refer to our
[public project status tracker](https://github.com/orgs/tensorflow/projects/4)
and