// CHECK: %[[REDUCED:.*]] = mhlo.reduce(%[[CAST]] init: %[[INITIAL]]) applies mhlo.maximum across dimensions = [1] : (tensor<4x8xf16>, tensor<f16>) -> tensor<4xf16>
  // CHECK: %[[CAST_BACK:.*]] = mhlo.convert %[[REDUCED]] : tensor<4xf16>
  // CHECK: %[[RESULT:.*]] = mhlo.reshape %[[CAST_BACK]] : (tensor<4xf16>) -> tensor<4x1xf16>
  // CHECK: return %[[RESULT]] : tensor<4x1xf16>

func.func @broadcast_to_i16_low_dim(%arg0: tensor<3xi16>, %arg1: tensor<2xi32>) -> tensor<3x3xi16> {
  %0 = "tfl.broadcast_to"(%arg0, %arg1) : (tensor<3xi16>, tensor<2xi32>) -> tensor<3x3xi16>
  return %0 : tensor<3x3xi16>
  // CHECK:  %cst = arith.constant dense<1> : tensor<3x3xi16>
  // CHECK:  %0 = tfl.mul(%arg0, %cst) <{fused_activation_function = "NONE"}> : (tensor<3xi16>, tensor<3x3xi16>) -> tensor<3x3xi16>

  func.return %4#0, %4#1 : tensor<4x32xf32>, tensor<4x32xi32>
}

// CHECK-LABEL:   func @convert_argmin_i16(
// CHECK-SAME:                         %[[VAL_0:.*]]: tensor<2xi16>) -> (tensor<i16>, tensor<i32>) {
// CHECK:           %[[VAL_9:.*]] = arith.constant dense<0> : tensor<1xi32>
// CHECK:           %[[VAL_10:.*]] = "tf.Min"(%[[VAL_0]], %[[VAL_9]]) <{keep_dims = false}> : {{.*}} -> tensor<i16>