27:127>:f32, 0.035427000373601913>>, %arg9: tensor<2048x!quant.uniform<i32:f32, 4.2675782196965883E-7>>, %arg10: tensor<2048x!quant.uniform<i32:f32, 1.0742187583900886E-7>>, %arg11: tensor<2048x!quant.uniform<i32:f32, 1.6406249869760359E-7>>, %arg12: tensor<2048x!quant.uniform<i32:f32, 1.523437447303877E-7>>, %arg13: tensor<640x2048x!quant.uniform<i8<-127:127>:f32, 0.021174000576138496>>, %arg14: tensor<640x!quant.uniform<i32:f32, 1.601389680352559E-4>>, %arg15: tensor<2048x!quant.uniform<i16:f32,...

  %k = "tf_device.launch"() ( {
    %9 = "tf.opK"(%input_0, %input_4, %9) :
      (tensor<i32>, tensor<!tf_type.variant>, tensor<i1>) -> tensor<i32>
    tf_device.return %9 : tensor<i32>
  }) {device = "DEVICE_ALIAS_0"} : () -> tensor<i32>
  %l = "tf_device.launch"() ( {
    %10 = "tf.opL"(%input_1, %input_4, %9) :
      (tensor<f32>, tensor<!tf_type.variant>, tensor<i1>) -> tensor<f32>

27:127>:f32, 0.035427000373601913>>, %arg9: tensor<2048x!quant.uniform<i32:f32, 4.2675782196965883E-7>>, %arg10: tensor<2048x!quant.uniform<i32:f32, 1.0742187583900886E-7>>, %arg11: tensor<2048x!quant.uniform<i32:f32, 1.6406249869760359E-7>>, %arg12: tensor<2048x!quant.uniform<i32:f32, 1.523437447303877E-7>>, %arg13: tensor<640x2048x!quant.uniform<i8<-127:127>:f32, 0.021174000576138496>>, %arg14: tensor<640x!quant.uniform<i32:f32, 1.601389680352559E-4>>, %arg15: tensor<2048x!quant.uniform<i16:f32,...

  %q_bias = "quantfork.qcast"(%bias) : (tensor<2xf32>) -> tensor<2x!quant.uniform<i32:f32, 0.044022349891595126>>
  %dq_bias = "quantfork.dcast"(%q_bias) : (tensor<2x!quant.uniform<i32:f32, 0.044022349891595126>>) -> tensor<2xf32>

  %result = tf_executor.graph {

// CHECK: tf_executor.Merge{{.*}}(tensor<*xf32>, tensor<4xf32>) -> (tensor<4xf32>, tensor<i32>, !tf_executor.control)
    %value, %idx, %ctlMerge = tf_executor.Merge %arg0, %arg1  : (tensor<*xf32>, tensor<4xf32>) -> (tensor<4xf32>, tensor<i32>, !tf_executor.control)
    tf_executor.fetch %value : tensor<4xf32>
  }
  func.return %result : tensor<4xf32>
}

  %loop.init, %ctl0 = tf_executor.Enter %count.init : i32

  %next_count, %tok = tf_executor.NextIteration.Source : i32

  %loop.body.init, %ctlMerge = tf_executor.Merge %loop.init, %next_count : i32

  %dec_count, %ctlAdd = tf_executor.island
    wraps tf.Add %loop.body.init, -1 : (i32, i32) -> i32

  %loop_cond, %ctlNE = tf_executor.island
    wraps tf.NotEqual %dec_count, 0 : (i32, i32) -> i1

func.func private @testIf1Else(tensor<*xf32>) -> tensor<*xf32>

// CHECK-LABEL: func @testIf1Result(%arg0: tensor<i32>, %arg1: tensor<*xf32>)
func.func @testIf1Result(%arg0: tensor<i32>, %arg1: tensor<*xf32>) -> tensor<*xf32> {
  %0 = "tf.If"(%arg0, %arg1) {
    then_branch = @testIf1Then, else_branch = @testIf1Else, is_stateless = false
  } : (tensor<i32>, tensor<*xf32>) -> tensor<*xf32>

  // CHECK: [[ToBool:%.*]] = "tf.ToBool"

// CHECK-DAG: %[[input_10:.*]] = "tfl.dequantize"({{.*}}) : (tensor<3x!quant.uniform<i32:f32, 2.4030322780124744E-8>>) -> tensor<3xf32>
// CHECK-DAG: %[[input_11:.*]] = "tfl.dequantize"({{.*}}) : (tensor<3x!quant.uniform<i32:f32, 4.8060645560249487E-8>>) -> tensor<3xf32>
// CHECK-DAG: %[[input_12:.*]] = "tfl.dequantize"({{.*}}) : (tensor<3x!quant.uniform<i32:f32, 7.2090970130772759E-8>>) -> tensor<3xf32>

  %3 = "tfl.pack"(%1, %2) {tac.device = "CPU", tac.inference_type = "FLOAT", axis = 0 : i32, values_count = 2 : i32} : (tensor<1xf32>, tensor<1xf32>) -> tensor<2x1xf32>
  return %3 : tensor<2x1xf32>
}
```

In this code, `%3` is annotated with "CPU", while others are annotated with

		t.Errorf("float64(u16) != du16")
	}
	if float64(u8) != du8 {
		t.Errorf("float64(u8) != du8")
	}
	if float64(i64) != di64 {
		t.Errorf("float64(i64) != di64")
	}
	if float64(i32) != di32 {
		t.Errorf("float64(i32) != di32")
	}
	if float64(i16) != di16 {
		t.Errorf("float64(i16) != di16")
	}
	if float64(i8) != di8 {
		t.Errorf("float64(i8) != di8")
	}
	if float32(u64) != su64 {