return %1 : tensor<1x3xf32>
}
// CHECK: @main(%[[ARG_0:.+]]: tensor<1x4xf32>) -> tensor<1x3xf32>
// CHECK-DAG: %[[CST:.+]] = stablehlo.constant dense<1.000000e+00> : tensor<4x3xf32>

  func.return %3 : tensor<2x1xf32>
}

// CHECK: %[[CST:.*]] = arith.constant dense<1> : tensor<4xi32>
// CHECK:  [[VAL_0:%.*]] = "tfl.reshape"(%1, %[[CST]]) {tac.device = "GPU",  tac.inference_type = "FLOAT"} : (tensor<1xf32>, tensor<4xi32>) -> tensor<1x1x1x1xf32>
// CHECK:  [[VAL_1:%.*]] = "tfl.reshape"(%2, %[[CST]]) {tac.device = "GPU",  tac.inference_type = "FLOAT"} : (tensor<1xf32>, tensor<4xi32>) -> tensor<1x1x1x1xf32>

    func.func @main(%arg0: tensor<1x2xf32>) -> tensor<1x3xf32> {
      %cst = "tf.Const"() {value = dense<3.00000000e-1> : tensor<2x3xf32>} : () -> tensor<2x3xf32>
      %0 = "quantfork.stats"(%arg0) {layerStats = dense<[6.00000000e-6, 9.00000000e-1]> : tensor<2xf32>} : (tensor<1x2xf32>) -> tensor<1x2xf32>

func.func @main(tensor<40x37xf32>, tensor<40x37xf32>) -> tensor<40x40xf32> {
^bb0(%arg0: tensor<40x37xf32>, %arg1: tensor<40x37xf32>):
  %cst = arith.constant dense<1.0> : tensor<40xf32>
  %0:2 = "tfl.fully_connected"(%arg0, %arg1, %cst) {fused_activation_function = "NONE", keep_num_dims = false, weights_format = "DEFAULT"} : (tensor<40x37xf32>, tensor<40x37xf32>, tensor<40xf32>) -> (tensor<40x40xf32>, tensor<40x40xf32>)

  %cst = arith.constant dense<1.000000e+00> : tensor<256x256xf32>
  %cst_0 = arith.constant dense<0> : tensor<i32>
  %cst_1 = arith.constant dense<-1> : tensor<i32>

  }
}

// CHECK-LABEL: quantize_dot_general_per_tensor
// CHECK-SAME: %[[ARG0:.+]]: tensor<1x2xf32>
// CHECK: %[[CST:.+]] = stablehlo.constant() <{value = dense<127> : tensor<2x3xi8>}> : () -> tensor<2x3x!quant.uniform<i8<-127:127>:f32, 0.0023622048182750312>>

  %cst = "tfl.pseudo_qconst"() {qtype = tensor<1x2x!quant.uniform<u8:f32, 1.0>>, value = dense<-76> : tensor<1x2xi8>} : () -> tensor<1x2x!quant.uniform<u8:f32, 1.0>>

    %cst = stablehlo.constant dense<0.000000e+00> : tensor<10x1x3xf32>
    %0 = stablehlo.dot_general %arg0, %arg1, batching_dims = [0] x [0], contracting_dims = [2] x [1], precision = [DEFAULT, DEFAULT] {mhlo.frontend_attributes = {grad_x = "false", grad_y = "false"}} : (tensor<10x1x1024xf32>, tensor<10x1024x3xf32>) -> tensor<10x1x3xf32>
    %1 = stablehlo.maximum %0, %cst : tensor<10x1x3xf32>
    return %1 : tensor<10x1x3xf32>

        self.assertIn('stablehlo.add %arg0, %cst : tensor<4xf32>', str(module))

  def test_tf_mlir_to_stablehlo(self):
    assembly = """
      module attributes {tf.versions = {bad_consumers = [], min_consumer = 0 : i32, producer = 268 : i32}} {
        func.func @main(%arg0 : tensor<?xf32>) -> tensor<?xf32> {
          %cst = "tf.Const"() {value = dense<1.0> : tensor<f32>} : () -> tensor<f32>

  %0 = "tf.BroadcastTo"(%arg0, %arg1) : (tensor<3xbf16>, tensor<2xi64>) -> tensor<3x3xbf16>
  func.return %0: tensor<3x3xbf16>

// CHECK-LABEL: broadcast_to_bf16
// CHECK:  [[CST:%.*]] = arith.constant dense<1.000000e+00> : tensor<3x3xbf16>
// CHECK:  [[MUL:%.*]] = tfl.mul(%arg0, [[CST]]) <{fused_activation_function = "NONE"}> : (tensor<3xbf16>, tensor<3x3xbf16>) -> tensor<3x3xbf16>
// CHECK:  return [[MUL]] : tensor<3x3xbf16>