module attributes {} {
      func.func @main(%arg0: tensor<1x4xf32>) -> tensor<1x3xf32> attributes {} {
        %0 = stablehlo.constant dense<1.0> : tensor<4x3xf32>
        %1 = stablehlo.dot_general %arg0, %0, contracting_dims = [1] x [0], precision = [DEFAULT, DEFAULT] : (tensor<1x4xf32>, tensor<4x3xf32>) -> tensor<1x3xf32>
        return %1 : tensor<1x3xf32>
      }
    }
  )mlir");
  ASSERT_TRUE(module_op);

  return %0 : tensor<1x3xf32>
}

// CHECK-LABEL: @_func
func.func private @_func(%arg0: tensor<1x3xf32>) -> (tensor<1x3xf32>) {
  return %arg0 : tensor<1x3xf32>
}
}


// -----
// SPMD: one input and no return
//
// CHECK-LABEL: func.func @serving_default(%arg0: tensor<1x3xf32>) {
// CHECK-NEXT:  "tf.IfrtCall"(%arg0)

  func.func @matmul(%arg0: tensor<2x12xf32>) -> (tensor<*xf32>) {
    %cst_0 = "tf.Const"() {value = dense<0.000000e+00> : tensor<12x2xf32>} : () -> tensor<12x2xf32>
    %1 = "tf.PartitionedCall"(%arg0, %cst_0) {_tfl_quant_trait = "fully_quantizable", config = "", config_proto = "", executor_type = "", f = @composite_matmul_fn_1} : (tensor<2x12xf32>, tensor<12x2xf32>) -> tensor<*xf32>
    func.return %1: tensor<*xf32>
  }

  %cst = "tf.Const"() {device = "", value = dense<1.0> : tensor<4x3x3xf32>} : () -> tensor<4x3x3xf32>
  %0 = "tf.BatchMatMulV2"(%arg0, %cst) {adj_x = false, adj_y = false, device = ""} : (tensor<4x4x3xf32>, tensor<4x3x3xf32>) -> tensor<4x4x3xf32>
  return %0 : tensor<4x4x3xf32>

// CHECK-DAG: %[[CST:.*]] = "tf.Const"() {{.*}} : () -> tensor<4x3x3xf32>
// CHECK: %[[PARTITIONEDCALL_0:.*]] = "tf.PartitionedCall"(%arg0, %[[CST]])

  %recurrent_input = "tfl.pseudo_const"() {value = dense<0.000000e+00> : tensor<1x3xf32>} : () -> tensor<1x3xf32>
  %cell_input = "tfl.pseudo_const"() {value = dense<1.000000e+00> : tensor<1x3xf32>} : () -> tensor<1x3xf32>
  %16 = "tfl.unidirectional_sequence_lstm"(
    %arg0,
    %1, %2, %3, %4,
    %5, %6, %7, %8,
    %9, %9, %9,
    %10, %11,
    %10, %10,

  %recurrent_input = "tfl.pseudo_const"() {value = dense<0.000000e+00> : tensor<1x3xf32>} : () -> tensor<1x3xf32>
  %recurrent_stats = "quantfork.stats"(%recurrent_input) {layerStats = dense<[0.0, 1.0]> : tensor<2xf32>} : (tensor<1x3xf32>) -> tensor<1x3xf32>
  %cell_input = "tfl.pseudo_const"() {value = dense<1.000000e+00> : tensor<1x3xf32>} : () -> tensor<1x3xf32>

// RUN: flatbuffer_translate -mlir-to-tflite-flatbuffer %s -o - | flatbuffer_to_string - | FileCheck %s

func.func @main(tensor<40x37xf32>, tensor<40x37xf32>) -> tensor<40x40xf32> {
^bb0(%arg0: tensor<40x37xf32>, %arg1: tensor<40x37xf32>):
  // CHECK:      {
  // CHECK-NEXT:  version: 3,
  // CHECK-NEXT:  operator_codes: [ {
  // CHECK-NEXT:    deprecated_builtin_code: 9,
  // CHECK-NEXT:    version: 1,
  // CHECK-NEXT:    builtin_code: FULLY_CONNECTED

// RUN: flatbuffer_translate -mlir-to-tflite-flatbuffer %s -o - | flatbuffer_to_string - | FileCheck %s

func.func @main(tensor<40x37xf32>, tensor<40x37xf32>) -> tensor<40x40xf32> {
^bb0(%arg0: tensor<40x37xf32>, %arg1: tensor<40x37xf32>):
  // CHECK:      {
  // CHECK-NEXT:  version: 3,
  // CHECK-NEXT:  operator_codes: [ {
  // CHECK-NEXT:    deprecated_builtin_code: 9,
  // CHECK-NEXT:    version: 2,
  // CHECK-NEXT:    builtin_code: FULLY_CONNECTED

  // MLIR @main function corresponds to the TF function "main_original".
  OwningOpRef<ModuleOp> module_op = ParseModuleOpString(R"mlir(
    func.func private @main(%arg: tensor<1x2xf32>) -> (tensor<1x2xf32>) attributes {tf._original_func_name = "main_original"} {
      return %arg : tensor<1x2xf32>
    }
  )mlir");
  ASSERT_TRUE(module_op);

  absl::flat_hash_map<FunctionName, FunctionAlias> function_aliases;

  %cst2 = arith.constant dense<3.0> : tensor<23x2xf32>
  %0 = "tf.Conv2D"(%arg0, %cst0) {T = "tfdtype$DT_FLOAT", data_format = "NHWC", dilations = [1, 2, 3, 1], padding = "SAME", strides = [1, 4, 5, 1]} : (tensor<1x112x112x3xf32>, tensor<1x3x3x2xf32>) -> tensor<1x28x23x2xf32>
  %1 = "tf.Mul"(%0, %cst2) : (tensor<1x28x23x2xf32>, tensor<23x2xf32>) -> tensor<1x28x23x2xf32>

  func.return %1 : tensor<1x28x23x2xf32>