// CHECK:     "tfl.yield"(%[[RES1]]) : (tensor<*xf32>) -> ()
// CHECK:   }) : (tensor<1xf32>) -> tensor<*xf32>
// CHECK:   return %0 : tensor<*xf32>
// CHECK: }
func.func @main(%arg0: tensor<1xf32>) -> tensor<*xf32> {
  %0 = "tf.While"(%arg0) {cond = @cond, body = @body, is_stateless = false} : (tensor<1xf32>) -> tensor<*xf32>
  func.return %0 : tensor<*xf32>
}

func.func @cond(%arg1: tensor<*xf32>) -> tensor<*xf32> {

func.func @const() {
  tf_executor.graph {
    %0:2 = tf_executor.island wraps "tf.Const"() {device = "TPU:0", name = "const", dtype = "tfdtype$DT_INT32", value = dense<[1, 2]> : tensor<2xi32>} : () -> tensor<2xi32>
    tf_executor.fetch
  }
  func.return

    // CHECK: %[[sum:.*]] = "tf.Add"(%[[v]], %[[bar]])
    // CHECK: return %[[sum]]
    %0 = "tf.ReadVariableOp"(%arg0) : (tensor<!tf_type.resource<tensor<10xf32>>>) -> tensor<10xf32>
    %1 = "tf.VarHandleOp"() {container = "", shared_name = "bar"} : () -> tensor<!tf_type.resource<tensor<1xf32>>>
    %2 = "tf.ReadVariableOp"(%1) : (tensor<!tf_type.resource<tensor<1xf32>>>) -> tensor<1xf32>

  %cst = arith.constant dense<[-1, 7]> : tensor<2xi32>
  %0 = "tf.TensorListReserve"(%cst, %arg0) : (tensor<2xi32>, tensor<i32>) -> tensor<!tf_type.variant<tensor<?x7xf32>>>
  %1 = "tf.TensorListGetItem"(%0, %arg1, %cst) : (tensor<!tf_type.variant<tensor<?x7xf32>>>, tensor<i32>, tensor<2xi32>) -> tensor<?x7xf32>
  func.return %1 : tensor<?x7xf32>

      } : (tensor<1xf32>, tensor<f32>, tensor<i32>) -> tensor<1x!tf_type.qint8>
      %1 = "tf.UniformDequantize"(%0, %scales, %zps) {
        quantization_axis = -1 : i64, quantization_min_val = -128 : i64, quantization_max_val = 127 : i64
      } : (tensor<1x!tf_type.qint8>, tensor<f32>, tensor<i32>) -> tensor<1xf32>
      func.return %1 : tensor<1xf32>
    }
  })mlir";

         strides = [1, 1, 1, 1]
       } : (tensor<1x28x28x64xf32>, tensor<4xi32>, tensor<1x28x28x64xf32>)
        -> tensor<1x1x64x64xf32>

  func.return %0 : tensor<1x1x64x64xf32>
}

// CHECK-LABEL: func @transposeConv2DBackpropFilter_f16
func.func @transposeConv2DBackpropFilter_f16(
  %input:        tensor<1x64x28x28xf16>,
  %filter_size:  tensor<4xi32>,
  %out_backprop: tensor<1x64x28x28xf16>
) -> tensor<1x1x64x64xf16> {

// CHECK-NEXT:    buffer: 2
// CHECK-NEXT:  } ],
// CHECK-NEXT:  signature_defs: [  ]
// CHECK-NEXT: }
func.func @main(%arg0 : tensor<!tf_type.variant<tensor<2xi32>>>) -> tensor<!tf_type.variant<tensor<2xi32>>> {
  func.return %arg0 : tensor<!tf_type.variant<tensor<2xi32>>>

>, tensor<i8>, tensor<f64>, tensor<!tf_type.resource>)

# CHECK-SUBTYPE: func @main(%arg0: tensor<10xi32>, %arg1: tensor<!tf_type.variant<tensor<10xf32>>>, %arg2: tensor<10xi8>, %arg3: tensor<10xf64>, %arg4: tensor<!tf_type.resource<tensor<10xi32>>>) -> (tensor<10xi32>, tensor<!tf_type.variant<tensor<10xf32>>>, tensor<10xi8>, tensor<10xf64>, tensor<!tf_type.resource<tensor<10xi32>>>)...

  # checking.
  #
  # Note: CHECK-DAG doesn't work with CHECK-SAME/CHECK-NEXT.

  # Check index paths for results.
  #
  # CHECK:      func {{@[a-zA-Z_0-9]+}}() -> (
  # CHECK-SAME:   tensor<1xf32> {tf_saved_model.index_path = []})
  # CHECK-SAME: attributes {{.*}} tf_saved_model.exported_names = ["f0000_single_return"]
  @tf.function(input_signature=[])
  def f0000_single_return(self):

// CHECK-NEXT: tfl.gather
// CHECK-NEXT: tfl.dequantize
}

// CHECK-LABEL: dequantize_pushdown_gather_with_reduction
func.func @dequantize_pushdown_gather_with_reduction(%arg0: tensor<2xi32>) -> tensor<2x2xf32> {
  %w = "tfl.pseudo_qconst"() {qtype = tensor<12x2x!quant.uniform<i8<-127:127>:f32, 1.000000e+00>>, value = dense<127> : tensor<12x2xi8>} : () -> tensor<12x2x!quant.uniform<i8<-127:127>:f32, 1.000000e+00>>