EXPECT_TRUE(params_->body_graph != nullptr);
  EXPECT_TRUE(params_->cond_graph != nullptr);

  EXPECT_EQ(params_->ninputs, 2);

  ASSERT_TRUE(params_->cond_inputs != nullptr);
  ASSERT_TRUE(params_->cond_inputs[0].oper != nullptr);
  EXPECT_TRUE(params_->cond_inputs[1].oper != nullptr);

  ASSERT_TRUE(params_->body_inputs != nullptr);
  EXPECT_TRUE(params_->body_inputs[0].oper != nullptr);

// WhileOp
//===----------------------------------------------------------------------===//

static LogicalResult VerifyWhileTypes(Operation *op, TypeRange cond_input,
                                      TypeRange body_input,
                                      TypeRange body_result,
                                      bool shape_invariant) {

    if (!CreateInput(inputs[i], cond_graph, StrCat("cond_input", i).c_str(),
                     &cond_inputs[i], status)) {
      break;
    }
    if (!CreateInput(inputs[i], body_graph, StrCat("body_input", i).c_str(),
                     &body_inputs[i], status)) {
      break;
    }
  }

  TF_WhileParams params = {ninputs,    cond_graph,  cond_inputs,  cond_output,

  func.func @f(%arg1: tensor<!tf_type.resource<tensor<f32>>> {tf_saved_model.bound_input = @"v"}, %arg2: tensor<!tf_type.resource<tensor<f32>>> {tf_saved_model.bound_input = @"v2"})
  attributes {tf_saved_model.exported_names = ["f"]} {
    // CHECK-DAG: "tf.Const"() <{value = dense<1.000000e+00> : tensor<f32>}>

  // CHECK: func @h(%arg0: tensor<!tf_type.resource<tensor<f32>>> {tf_saved_model.bound_input = @c})
  func.func @h(%arg0: tensor<!tf_type.resource<tensor<f32>>> {tf_saved_model.bound_input = @c})
  attributes {tf_saved_model.exported_names = ["h"]} {
    %0 = "tf.ReadVariableOp"(%arg0) : (tensor<!tf_type.resource<tensor<f32>>>) -> tensor<f32>
    func.return
  }

  // CHECK: func @f(%arg0: tensor<!tf_type.resource<tensor<f32>>> {tf_saved_model.bound_input = @v})
  func.func @f(%arg0: tensor<!tf_type.resource<tensor<f32>>> {tf_saved_model.bound_input = @v}) -> (tensor<f32> {tf_saved_model.index_path = []})
  attributes {tf_saved_model.exported_names = ["f"]} {

  // CHECK-LABEL: @read_from_global
  func.func @read_from_global(%arg0: tensor<!tf_type.resource<tensor<10xf32>>> {tf_saved_model.bound_input = @v})
  attributes {tf_saved_model.exported_names = ["read_from_global"]} {
    // CHECK: [[name:%.*]] = "tf.VarHandleOp"
    // CHECK: "tf.ReadVariableOp"([[name]])

# CHECK:      func @[[init]]
# CHECK-SAME: [[ARG0:%.*]]: tensor<!tf_type.string> {tf_saved_model.bound_input = @[[asset0]]}
# CHECK-SAME: [[ARG1:%.*]]: tensor<!tf_type.string> {tf_saved_model.bound_input = @[[asset1]]}
# CHECK-SAME: tf_saved_model.initializer_type = "init_op"
# CHECK-NEXT: [[R0:%.*]] = "tf.HashTableV2"()
# CHECK-SAME: shared_name = "[[hash_table:.*]]"

    Operation *op, unsigned region_index, unsigned arg_index,
    NamedAttribute named_attr) {
  if (named_attr.getName() == "tf_saved_model.bound_input") {
    if (!mlir::isa<FlatSymbolRefAttr>(named_attr.getValue())) {
      return op->emitError() << "'tf_saved_model.bound_input' attribute should "
                                "be a FlatSymbolRefAttr";
    }
    auto symbol_name =

    module {
      "tf_saved_model.asset"() {filename = "assets/file_0.txt", sym_name = "__tf_saved_model_asset0"} : () -> ()
      func.func @main(%arg_0: tensor<!tf_type.string> {tf_saved_model.bound_input = @__tf_saved_model_asset0}) -> () attributes {tf.entry_function = {inputs = "arg_0:0", outputs = ""}} {
        return
      }
    }
  )mlir");

  FailureOr<SmallVector<AssetFileDef>> asset_file_defs =