%3 = "tf.XlaPad"(%arg0, %arg1, %0, %1, %2) : (tensor<2x2xf64>, tensor<f64>, tensor<2xi32>, tensor<2xi32>, tensor<2xi32>) -> tensor<6x5xf64>
    func.return %3 : tensor<6x5xf64>
  }

  func.func @non_const_inputs(%arg0: tensor<2x2xf64>, %arg1: tensor<f64>, %arg2: tensor<2xi32>, %arg3: tensor<2xi32>, %arg4: tensor<2xi32>) -> tensor<6x5xf64> {

// CHECK-LABEL: @add_f64(%arg0: tensor<3x3xbf16>, %arg1: tensor<3x3xbf16>) -> tensor<3x3xbf16>
func.func @add_f64(%arg0: tensor<3x3xf64>, %arg1: tensor<3x3xf64>) -> tensor<3x3xf64> {
  // CHECK-NOT: f64
  // CHECK: stablehlo.add
  %0 = stablehlo.add %arg0, %arg1: (tensor<3x3xf64>, tensor<3x3xf64>) -> tensor<3x3xf64>
  return %0 : tensor<3x3xf64>
}

// -----

// CHECK-LABEL: @constant_f32() -> tensor<2x2xbf16>

  // CHECK: %[[RESHAPE:.*]] = stablehlo.reshape
  // CHECK-NEXT: return %[[RESHAPE]]

  // CHECK: @main_1
  func.func private @main_1(%arg0: tensor<3x10xf32>) -> tensor<6x5xf32> {
    %0 = stablehlo.reshape %arg0 : (tensor<3x10xf32>) -> tensor<6x5xf32>
    return %0 : tensor<6x5xf32>
  }
  // CHECK: %[[RESHAPE:.*]] = stablehlo.reshape
  // CHECK-NEXT: return %[[RESHAPE]]

  // CHECK: %[[RHS_2:.*]] = "tf.Reshape"(%[[RHS_SPLIT]]#1, %[[MATMUL_RHS_SHAPE]]) : (tensor<1x6x5xf32>, tensor<2xi64>) -> tensor<6x5xf32>
  // CHECK: %[[RHS_3:.*]] = "tf.Reshape"(%[[RHS_SPLIT]]#2, %[[MATMUL_RHS_SHAPE]]) : (tensor<1x6x5xf32>, tensor<2xi64>) -> tensor<6x5xf32>

func.func @broadcast_splat_operand() -> tensor<5x5xi64> {
  %cst = arith.constant dense<5> : tensor<2xi64>
  %0 = "tf.BroadcastTo"(%cst, %cst) : (tensor<2xi64>, tensor<2xi64>) -> tensor<5x5xi64>
  func.return %0 : tensor<5x5xi64>
  // CHECK: %[[V0:.*]] = "tf.Const"() <{value = dense<5> : tensor<5x5xi64>}> : () -> tensor<5x5xi64>
  // CHECK: %[[V0]] : tensor<5x5xi64>

        # CHECK: value = dense<{{\[\[}}1.000000e+00, 2.000000e+00, 3.000000e+00], [4.000000e+00, 5.000000e+00, 6.000000e+00]]> : tensor<2x3xf64>
      }
    }
  }
}
node {
  name: "x"
  op: "Const"
  device: "/device:CPU:0"
  attr {
    key: "dtype"
    value {
      type: DT_FLOAT
    }
  }
  attr {
    key: "value"

  %arg4: tensor<!tf_type.resource<tensor<64xf32>>>, %arg5: tensor<!tf_type.resource<tensor<128xf32>>>)
    -> (tensor<i64>, tensor<i32>) attributes {tf._tf_data_function = true, tf.signature.is_stateful} {

}

// CHECK-LABEL: func @testEmptyFloat
func.func @testEmptyFloat() -> (tensor<5xf64>) {
  %0 = "tf.Const"() { value = dense<5> : tensor<i32> } : () -> tensor<i32>

  // CHECK: [[VAL:%.+]] = "tf.Const"() <{value =  dense<0.000000e+00> : tensor<5xf64>}>
  // CHECK: return [[VAL]]
  %1 = "tf.Empty"(%0) : (tensor<i32>) -> (tensor<5xf64>)
  func.return %1 : tensor<5xf64>
}

// CHECK-LABEL: func @testEmptyf16

  // CHECK-DAG: %[[LE2:.*]] = "tf.LessEqual"(%[[OFFSET]], %[[NUM_LOWER_OR_M]]) : (tensor<4x5xi64>, tensor<i64>) -> tensor<4x5xi1>
  // CHECK-DAG: %[[NEG:.*]] = "tf.Neg"(%[[NUM_UPPER_OR_N]]) : (tensor<i64>) -> tensor<i64>
  // CHECK-DAG: %[[GE:.*]] = "tf.GreaterEqual"(%[[OFFSET]], %[[NEG]]) : (tensor<4x5xi64>, tensor<i64>) -> tensor<4x5xi1>

// `tfl.dequantize`.

func.func @uniform_dequantize_op_return_f64(%arg: tensor<2x2x!quant.uniform<i8:f64, 1.000000e+0:8>>) -> tensor<2x2xf64> {
  %0 = stablehlo.uniform_dequantize %arg : (tensor<2x2x!quant.uniform<i8:f64, 1.000000e+0:8>>) -> tensor<2x2xf64>
  return %0 : tensor<2x2xf64>
}
// CHECK-LABEL: uniform_dequantize_op_return_f64
// CHECK: stablehlo.uniform_dequantize
// CHECK-NOT: tfl.dequantize

// -----