%1 = "tfl.pseudo_const"() { value = dense<1.0> : tensor<16xf32>} : () -> tensor<16xf32>
  %2 = "tfl.conv_3d"(%arg0, %0, %1) {dilation_h_factor = 1 : i32, dilation_w_factor = 1 : i32, dilation_d_factor = 1 : i32, fused_activation_function = "NONE", padding = "SAME", stride_h = 1 : i32, stride_w = 1 : i32, stride_d = 1 : i32} : (tensor<1x32x32x32x8xf32>, tensor<1x1x1x8x16x!quant.uniform<i8:f32, 1.0>>, tensor<16xf32>) -> tensor<1x32x32x32x16xf32>

  %1 = "tfl.sum"(%0, %cst) {keep_dims = false} : (tensor<2xf32>, tensor<1xi32>) -> tensor<f32>
  %2 = "tfl.add"(%1, %cst_1) {fused_activation_function = "NONE"} : (tensor<f32>, tensor<1xf32>) -> tensor<1xf32>
  %3 = "tfl.rsqrt"(%2) : (tensor<1xf32>) -> tensor<1xf32>
  %4 = "tfl.mul"(%arg0, %3) {fused_activation_function = "NONE"} : (tensor<2xf32>, tensor<1xf32>) -> tensor<2xf32>
  func.return %4: tensor<2xf32>

  // Use other output
  %3:6 = "tf.FusedBatchNormV3"( %2#0, %arg1, %arg2, %arg3, %arg4) {T = "tfdtype$DT_FLOAT", U = "tfdtype$DT_FLOAT", data_format = "NHWC", epsilon = 0.001 : f32, is_training = false} : (tensor<8x8x8x8xf32>, tensor<8xf32>, tensor<8xf32>, tensor<8xf32>, tensor<8xf32>) -> (tensor<8x8x8x8xf32>, tensor<8xf32>, tensor<8xf32>, tensor<8xf32>, tensor<8xf32>, tensor<8xf32>)

        tensor<20xf32>, tensor<20xf32>, tensor<20xf32>,
        tensor<20xf32>, tensor<20xf32>, tensor<20xf32>, tensor<20xf32>,
        tensor<20x20xf32>, none,
        tensor<1x20xf32>, tensor<1x20xf32>,
        none, none, none, none) -> tensor<1x28x20xf32>
    %1 = "quantfork.stats"(%0) {layerStats = dense<[-1.0, 2.0]> : tensor<2xf32>} : (tensor<1x28x20xf32>) -> tensor<1x28x20xf32>

}

func.func @div(%arg0: tensor<1xf32>, %arg1: tensor<1xf32>) -> tensor<1xf32> {
  %0 = "tf.Div"(%arg0, %arg1) : (tensor<1xf32>, tensor<1xf32>) -> tensor<1xf32>
  func.return %0: tensor<1xf32>

// CHECK-LABEL: div
// CHECK:  tfl.div %arg0, %arg1 {fused_activation_function = "NONE"} : tensor<1xf32>
// CHECK:  return
}

func.func @squaredDifferenceRelu(tensor<1xf32>, tensor<1xf32>) -> tensor<1xf32> {

  %0 = stablehlo.constant dense<2.000000e+00> : tensor<167x64xf32>
  %1 = stablehlo.constant dense<2.000000e+00> : tensor<64xf32>
  %2 = stablehlo.dot_general %arg0, %0, contracting_dims = [2] x [0], precision = [DEFAULT, DEFAULT] : (tensor<1x1x167xf32>, tensor<167x64xf32>) -> tensor<1x1x64xf32>
  %3 = stablehlo.broadcast_in_dim %1, dims = [2] : (tensor<64xf32>) -> tensor<1x1x64xf32>
  %4 = stablehlo.add %2, %3 : tensor<1x1x64xf32>
  func.return %4: tensor<1x1x64xf32>
}

  func.return %0#0 : tensor<?x?x?x?xf32>
}

// -----

// CHECK-LABEL: fusedBatchNormV3_training_dynamic_unsupported1

  %outputs_50 =  "tf.Mul"(%outputs_30, %outputs_48) {device = ""} : (tensor<*xf32>, tensor<*xf32>) -> tensor<*xf32>
  %outputs_52 =  "tf.Reshape"(%outputs_50, %outputs_0) {device = ""} : (tensor<*xf32>, tensor<2xi32>) -> tensor<*xf32>
  %outputs_54 =  "tf.MatMul"(%outputs_40, %outputs_52) {device = "", transpose_a = false, transpose_b = false} : (tensor<*xf32>, tensor<*xf32>) -> tensor<*xf32>

  %0 = "tfl.unidirectional_sequence_lstm"(%arg0,...

  %2 = "tf.Mul"(%1, %value) {T = "tfdtype$DT_FLOAT"} : (tensor<256x8x7x16xf32>, tensor<16xf32>) -> tensor<256x8x7x16xf32>
  func.return %2 : tensor<256x8x7x16xf32>

// CHECK-DAG: %[[cst:.*]] = "tf.Const{{.*}} dense<8.000000e+00> : tensor<3x3x3x16xf32>
// CHECK-DAG: %[[cst_0:.*]] = "tf.Const{{.*}} dense<1.200000e+01> : tensor<16xf32>