}

func.func @reduce_min(%arg0: tensor<8x16x16xf32>, %arg1: tensor<2xi32>) -> tensor<?xf32> {
  %0 = "tf.Min"(%arg0, %arg1) {keep_dims = false} : (tensor<8x16x16xf32>, tensor<2xi32>) -> tensor<?xf32>
  func.return %0 : tensor<?xf32>

  // CHECK-LABEL: reduce_min
  // CHECK: "tfl.reduce_min"(%arg0, %arg1) <{keep_dims = false}> : (tensor<8x16x16xf32>, tensor<2xi32>) -> tensor<?xf32>

  let results = (outs
    TFL_TensorOf<[F32, I32, I64, QI8, QUI8, TFL_Quint8, QI16]>:$output);

  let hasOptions = 1;
  let customOption = "ReducerOptions";
}

def TFL_ReduceMaxOp: TFL_Op<"reduce_max", [
    PredOpTrait<"input and output must have same element type",
      TFL_TCresVTEtIsSameAsOp<0, 0>>,
    Pure,
    QuantizableResult,
    SameOperandsAndResultsScale]> {

values.

```python

  def stable_softmax(x):
    z = x - tf.reduce_max(x)
    numerator = tf.exp(z)
    denominator = tf.reduce_sum(numerator)
    return numerator / denominator
```

However, when we backprop through the softmax to x, we dont want to backprop
through the `tf.reduce_max(x)` (if the max values are not unique then the

`tf.reduce_join`: `reduction_indices` becomes `axis` * `tf.reduce_logsumexp`:
`reduction_indices` becomes `axis` * `tf.reduce_max`: `reduction_indices`
becomes `axis` * `tf.reduce_mean`: `reduction_indices` becomes `axis` *
`tf.reduce_min`: `reduction_indices` becomes `axis` * `tf.reduce_prod`:
`reduction_indices` becomes `axis` * `tf.reduce_sum`: `reduction_indices`