// -----

// CHECK-LABEL: maxpool_explicit_padding
func.func @maxpool_explicit_padding(%arg0: tensor<2x12x20x7xi32>) -> tensor<2x3x5x7xi32> {
  // CHECK: tf.MaxPool
  // TODO(b/165938852): need to support explicit padding in max_pool.

  %0 = "tf.MaxPool"(%arg0) {data_format = "NHWC", ksize = [1, 2, 2, 1], padding = "EXPLICIT", strides = [1, 4, 4, 1]} : (tensor<2x12x20x7xi32>) -> tensor<2x3x5x7xi32>

}

// CHECK-LABEL:   func @convert_maxpool_valid(
// CHECK-SAME:                                %[[VAL_0:.*]]: tensor<4x16x16x8xf32>) -> tensor<4x7x7x8xf32> {
// CHECK:           %[[VAL_1:.*]] = "tf.MaxPool"(%[[VAL_0]]) <{data_format = "NHWC", explicit_paddings = [], ksize = [1, 3, 3, 1], padding = "VALID", strides = [1, 2, 2, 1]}> : (tensor<4x16x16x8xf32>) -> tensor<4x7x7x8xf32>
// CHECK:           return %[[VAL_1]] : tensor<4x7x7x8xf32>

};

using ConvertAvgPool2DGradOp =
    ConvertAvgPoolGradOp<TF::AvgPoolGradOp, /*num_dims=*/4>;
using ConvertAvgPool3DGradOp =
    ConvertAvgPoolGradOp<TF::AvgPool3DGradOp, /*num_dims=*/5>;

// Converts MaxPool op to HLO ReduceWindow op by setting appropriate window
// dimensions with max as the reduction function.
//
// Sample result for VALID padding mode:
//
//   %init = arith.constant dense<...> : tensor<i32>