<< x_name
          << "\nNumber of window dimensions: " << window_dimensions.size()
          << "\nNumber of " << x_name << ": " << x << "\n";
      return false;
    }
  };

  if (!(verify_size(window_dimensions.size(), "window_dimensions") &&
        verify_size(window_strides.size(), "window strides") &&
        verify_size(padding.size(), "padding entries") &&

    for (int64_t i = 0, e = window_dimensions.size(); i < e; ++i) {
      int64_t window_dimension = window_dimensions.getValues<int64_t>()[i];
      if (window_dimension == 1) continue;
      // Cumulative axis already set.
      if (cumulative_axis != -1) return failure();
      // Potential cumulative axis is not the right size.
      if (window_dimension != operand_type.getShape()[i]) return failure();

    %1 = stablehlo.maximum %arg2, %arg3 : tensor<!quant.uniform<i8:f32, 3.000000e-01:-5>>
    stablehlo.return %1 : tensor<!quant.uniform<i8:f32, 3.000000e-01:-5>>
  }) {window_dimensions = array<i64: 1, 3, 4, 1>, window_strides = array<i64: 1, 2, 3, 1>} : (tensor<2x9x10x3x!quant.uniform<i8:f32, 3.000000e-01:-5>>, tensor<!quant.uniform<i8:f32, 3.000000e-01:-5>>) -> tensor<2x4x3x3x!quant.uniform<i8:f32, 3.000000e-01:-5>>

  uint32_t opcode_index =
      GetOpcodeIndex(op_name, tflite::BuiltinOperator_STABLEHLO_REDUCE_WINDOW);

  auto wd = reduce_window_op.getWindowDimensions();
  auto window_dimensions =
      builder_.CreateVector(std::vector<int64_t>(wd.begin(), wd.end()));
  auto window_strides = builder_.CreateVector(
      mlir::GetOptionalVector<int64_t>(reduce_window_op.getWindowStrides()));

                                  << size << " vs. " << input_rank << ")";
        }
      }
    }
    return success();
  };

  if (check(op.getWindowDimensions(), "window_dimensions").failed())
    return failure();

  if (check(op.getWindowStrides(), "window_strides").failed()) return failure();

  if (check(op.getBaseDilations(), "base_dilations").failed()) return failure();

    if (op.getBaseDilations().has_value() ||
        op.getWindowDilations().has_value() || op.getPadding().has_value()) {
      return failure();
    }

    // Window_dimensions and window_strides should have batch and channel
    // dimension of 1 as they cannot be specified in tfl.max_pool_2d.
    ArrayRef<int64_t> window_dims = op.getWindowDimensions();

  %cst_1 = "tf.Const"() {value = dense<[2, 3, 1]> : tensor<3xi32>} : () -> tensor<3xi32>
  // expected-error @+1 {{'tf.XlaSelectAndScatter' op expects the size of window_dimensionsto be equal to the input rank (3 vs. 4)}}