// CHECK-SAME: %[[ARG3:[^:[:space:]]+]]
  func.func private @main(%arg0: tensor<1x4xf32>, %arg1: tensor<4x3xf32>) -> tensor<1x3xf32> attributes {_from_xla_call_module} {
    %0 = stablehlo.dot_general %arg0, %arg1, contracting_dims = [1] x [0] : (tensor<1x4xf32>, tensor<4x3xf32>) -> tensor<1x3xf32>
    return %0 : tensor<1x3xf32>
    // CHECK: %[[DOT:.+]] = stablehlo.dot_general %[[ARG2]], %[[ARG3]]
    // CHECK: return %[[DOT]]
  }
}

  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)

    return %2 : tensor<1x3xf32>
  }
  // CHECK-LABEL: func.func private @composite_dot_general_fn_1
  // CHECK-SAME: -> tensor<1x3xf32>
  func.func private @composite_dot_general_fn_1(%arg0: tensor<1x1024xf32>, %arg1: tensor<1024x3xf32>) -> tensor<1x3xf32> attributes {_from_xla_call_module} {

    return %3 : tensor<1x3xf32>
  }
  func.func private @composite_matmul_with_bias_fn_2(%arg0: tensor<1x4xf32>, %arg1: tensor<4x3xf32>, %arg2: tensor<3xf32>) -> tensor<1x3xf32> attributes {tf_quant.composite_function} {
    %0 = "tf.MatMul"(%arg0, %arg1) <{grad_a = false, grad_b = false, transpose_a = false, transpose_b = false}> {attr_map = "0:transpose_a,1:transpose_b", device = ""} : (tensor<1x4xf32>, tensor<4x3xf32>) -> tensor<1x3xf32>

func.func @pushTposeAfterAddSimpleWithFold(%arg0: tensor<2x3xi32>) -> tensor<3x2xi32> {
  %perm = arith.constant dense<[1, 0]> : tensor<2xi32>
  %0 = "tfl.transpose"(%arg0, %perm) : (tensor<2x3xi32>, tensor<2xi32>) -> tensor<3x2xi32>
  %cst = arith.constant dense<[[1, 2], [3, 4], [5, 6]]> : tensor<3x2xi32>
  %1 = tfl.add %0, %cst { fused_activation_function = "NONE" } : tensor<3x2xi32>
  func.return %1 : tensor<3x2xi32>
}

  %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>

// CHECK: return

  func.func private @composite_dot_general_fn(%arg0: tensor<1x4xf32>, %arg1: tensor<4x3xf32>) -> tensor<1x3xf32> attributes {_from_xla_call_module} {
      %0 = stablehlo.dot_general %arg0, %arg1, contracting_dims = [1] x [0] : (tensor<1x4xf32>, tensor<4x3xf32>) -> tensor<1x3xf32>
      return %0 : tensor<1x3xf32>
  }
}

// -----

func.func @dequantize_pushdown(%arg0: tensor<1000x1000x!quant.uniform<i8:f32, 7.812500e-03>>, %arg1: tensor<1x1xi32>) -> tensor<1x1x1000xf32> {
  %0 = "tfl.dequantize"(%arg0) : (tensor<1000x1000x!quant.uniform<i8:f32, 7.812500e-03>>) -> tensor<1000x1000xf32>
  %1 = "tfl.gather"(%0, %arg1) {axis = 0 : i32, batch_dims = 0 : i32}: (tensor<1000x1000xf32>, tensor<1x1xi32>) -> tensor<1x1x1000xf32>
  func.return %1 : tensor<1x1x1000xf32>

// CHECK-NEXT: tfl.gather

// CHECK-NEXT:    "tf.IfrtCall"(%arg0, [[ARRAYKEY]]) <{program_id = 6515870160938153680 : i64, variable_arg_indices = [1 : i32]}> {__tpu_compile_metadata_text = "retvals { sharding { } }"} : (tensor<1x3xf32>, tensor<!tf_type.string>) -> tensor<1x1xf32>
// CHECK-NEXT:    return
//
 func.func @serving_default(%arg0: tensor<1x3xf32>) -> tensor<1x1xf32> {