Output const0 = ops::Const(root.WithOpName("const0"), 1);
  Output const1 = ops::Const(root.WithOpName("const1"), 2);

  Output add = ops::Add(root.WithOpName("add"), const0, const1);

  root.graph()->AddControlEdge(id0.node(), const0.node());
  root.graph()->AddControlEdge(id1.node(), const1.node());

  std::unique_ptr<DeadnessAnalysis> result;

    TF_Operation* const3;
    if (grad_inputs_provided) {
      const float const3_val[] = {1.0, 1.0, 1.0, 1.0};
      const3 = FloatConst2x2(expected_graph_, s_, const3_val, "GradInputs");
    } else {
      const3 = OnesLike(expected_graph_, s_, matmul, "gradients/OnesLike");
    }

    TF_Operation* matmul1 = MatMul(expected_graph_, s_, const3, const1,

  // done in parallel.
  //
  // This graph is:
  // (Const0, Const0) -> MatMul0
  // (Const1, Const1) -> MatMul1
  // (MatMul0, MatMul1) -> MatMulCombined
  //
  // Device0: [Const0, Const0, MatMul0]
  // Device1: [Const1, Const1, MatMul1, MatMulCombined]
  //
  // Cluster0: [Const0, Const0, MatMul0]
  // Cluster1: [Const1, Const1, MatMul1]
  // Cluster2: [MatMulCombined]

	for {
		x := v_0
		s := v_1
		v.reset(OpInt64Make)
		v0 := b.NewValue0(v.Pos, OpRsh32x32, typ.UInt32)

  // CHECK: %[[CONST1:.*]] = "tf.Const"() <{value = dense<1> : tensor<i32>}> : () -> tensor<i32>
  %const1 = "tf.Const"() {value = dense<1> : tensor<i32>} : () -> tensor<i32>
  // CHECK: %[[SUB:.*]] = "tf.Sub"(%[[BARG1]], %[[CONST1]])
  %sub = "tf.Sub"(%arg1, %const1) : (tensor<i32>, tensor<i32>) -> tensor<i32>
  %elem = "tf._SomeOp"() : () -> tensor<3xf32>

  func.func private @quantize_dot_general_with_bias_same_shape_fn(%arg0: tensor<1x2xf32>) -> tensor<1x3xf32> attributes {tf._original_func_name = "main_0"} {
    %cst = "tf.Const"() {value = dense<3.00000000e-1> : tensor<2x3xf32>} : () -> tensor<2x3xf32>
    %cst_0 = "tf.Const"() {value = dense<4.00000000e-1> : tensor<1x3xf32>} : () -> tensor<1x3xf32>

// CHECK-LABEL: func @dynamic_shape_batch_matmul
// CHECK-DAG: %[[CONST:.*]] = "tf.Const"() <{value = dense<0> : tensor<1xi32>}> : () -> tensor<1xi32>
// CHECK-DAG: %[[CONST_1:.*]] = "tf.Const"() <{value = dense<2> : tensor<1xi32>}> : () -> tensor<1xi32>
// CHECK-DAG: %[[CONST_2:.*]] = "tf.Const"() <{value = dense<1> : tensor<1xi32>}> : () -> tensor<1xi32>
// CHECK-DAG: %[[CONST_3:.*]] = "tf.Const"() <{value = dense<[1024, 3]> : tensor<2xi64>}> : () -> tensor<2xi64>

			}
			p.advance(_Import, _Const, _Type, _Var, _Func)
			continue
		}

		// Reset p.pragma BEFORE advancing to the next token (consuming ';')
		// since comments before may set pragmas for the next function decl.
		p.clearPragma()

		if p.tok != _EOF && !p.got(_Semi) {
			p.syntaxError("after top level declaration")
			p.advance(_Import, _Const, _Type, _Var, _Func)
		}
	}
	// p.tok == _EOF

  // operations specified (0).
  void DefineT(int num_opers, const std::vector<TF_Operation*>& opers,
               const std::vector<TF_Output>& inputs,
               const std::vector<TF_Output>& outputs,
               const std::vector<string>& output_names,
               bool expect_failure = false) {
    ASSERT_EQ(func_, nullptr);
    const char** output_names_ptr = ToArray(output_names);

    tf_device.return %identity : tensor<i32>
  }) : () -> (tensor<i32>)
  return %cluster : tensor<i32>
}
```

will be transformed into:

```mlir
func @cluster() -> tensor<i32> {
  %cluster = "tf_device.cluster"() ( {
    %const = "tf.Const"() {value = dense<0> : tensor<i32>} : () -> tensor<i32>
    %identity = "tf.Identity"(%const) : (tensor<i32>) -> tensor<i32>