ASSERT_EQ(TF_OK, TF_GetCode(status)) << TF_Message(status);
  auto* h1_task1 =
      TFE_TensorHandleCopyToDevice(h1_task0, ctx, remote_device_name, status);
  ASSERT_EQ(TF_OK, TF_GetCode(status)) << TF_Message(status);

  TFE_Op* matmul = MatMulOp(ctx, h0_task1, h1_task1);
  TFE_OpSetDevice(matmul, remote_device_name, status);
  EXPECT_EQ(TF_OK, TF_GetCode(status)) << TF_Message(status);

  TFE_TensorHandle* retvals[1];

  ASSERT_TRUE(TF_GetCode(status.get()) == TF_OK) << TF_Message(status.get());
  bool arrived = false;
  bool executed = false;
  const char* name = "/job:localhost/replica:0/task:0/device:CUSTOM:0";
  RegisterLoggingDevice(context, name, /*strict_scope_placement=*/true,
                        &arrived, &executed, status.get());
  ASSERT_TRUE(TF_GetCode(status.get()) == TF_OK) << TF_Message(status.get());

                              status);
    CHECK_EQ(TF_GetCode(status), TF_OK) << TF_Message(status);

    matmul = TFE_NewOp(ctx, "MatMulFunction", status);
    ASSERT_EQ(TF_GetCode(status), TF_OK) << TF_Message(status);
    TFE_OpAddInput(matmul, h0_task0, status);
    ASSERT_EQ(TF_GetCode(status), TF_OK) << TF_Message(status);
    TFE_OpAddInput(matmul, has_packed_input ? packed_handle : h1_task2, status);

  TFE_Context* ctx = TFE_NewContext(opts, status);
  CHECK_EQ(TF_OK, TF_GetCode(status)) << TF_Message(status);
  TFE_DeleteContextOptions(opts);

  TFE_TensorHandle* h = TestScalarTensorHandle(ctx, 1.0f);
  TFE_TensorDebugInfo* debug_info = TFE_TensorHandleTensorDebugInfo(h, status);
  CHECK_EQ(TF_OK, TF_GetCode(status)) << TF_Message(status);

  ASSERT_EQ(0, TFE_TensorDebugInfoOnDeviceNumDims(debug_info));

    ASSERT_EQ(TF_OK, TF_GetCode(s)) << TF_Message(s);
    TF_OutputListPushBack(func_outputs, add_output2, s);
    ASSERT_EQ(TF_OK, TF_GetCode(s)) << TF_Message(s);
    TF_OutputListPushBack(func_outputs, mm_output, s);
    ASSERT_EQ(TF_OK, TF_GetCode(s)) << TF_Message(s);
    func = TF_FinalizeFunction(graph_ctx, func_outputs, s);
    ASSERT_EQ(TF_OK, TF_GetCode(s)) << TF_Message(s);
    TF_DeleteOutputList(func_outputs);
  }

  ASSERT_TRUE(TF_GetCode(status) == TF_OK) << TF_Message(status);
  TF_DeleteGraph(function_graph);
  TFE_ContextOptions* opts = TFE_NewContextOptions();
  TFE_Context* ctx = TFE_NewContext(opts, status);
  ASSERT_TRUE(TF_GetCode(status) == TF_OK) << TF_Message(status);
  TFE_DeleteContextOptions(opts);
  TFE_ContextAddFunction(ctx, fn, status);
  ASSERT_TRUE(TF_GetCode(status) == TF_OK) << TF_Message(status);
  TF_DeleteFunction(fn);

  EXPECT_EQ(TF_OK, TF_GetCode(status)) << TF_Message(status);

  TFE_TensorHandle* retvals[1];
  int num_retvals = 1;
  TFE_Execute(matmul, &retvals[0], &num_retvals, status);
  EXPECT_EQ(TF_OK, TF_GetCode(status)) << TF_Message(status);

  auto* retval_task0 =
      TFE_TensorHandleCopyToDevice(retvals[0], ctx, local_device_name, status);
  ASSERT_EQ(TF_OK, TF_GetCode(status)) << TF_Message(status);

  TFE_Op* op = TFE_NewOp(ctx, "AddV2", status);
  CHECK_EQ(TF_OK, TF_GetCode(status)) << TF_Message(status);
  TFE_OpAddInput(op, a, status);
  CHECK_EQ(TF_OK, TF_GetCode(status)) << TF_Message(status);
  TFE_OpAddInput(op, b, status);
  CHECK_EQ(TF_OK, TF_GetCode(status)) << TF_Message(status);
  TF_DeleteStatus(status);
  TFE_OpSetAttrType(op, "T", TFE_TensorHandleDataType(a));

  return op;
}

  TF_Tensor* tensor_1X6 = Int32Tensor({1, 6});
  CHECK_EQ(TF_OK, TF_GetCode(status_)) << TF_Message(status_);
  TF_Tensor* tensor_1X1X6 = Int32Tensor({1, 1, 6});
  CHECK_EQ(TF_OK, TF_GetCode(status_)) << TF_Message(status_);

  TFE_Op* reshape_op = TFE_NewOp(tfe_context_, "Reshape", status_);
  CHECK_EQ(TF_OK, TF_GetCode(status_)) << TF_Message(status_);
  TFE_OpSetAttrType(reshape_op, "T", TF_FLOAT);

    CHECK_EQ(TF_OK, TF_GetCode(status)) << TF_Message(status);
    TFE_OpAddInput(matmul, m, status);
    CHECK_EQ(TF_OK, TF_GetCode(status)) << TF_Message(status);
    TFE_OpAddInput(matmul, m, status);
    CHECK_EQ(TF_OK, TF_GetCode(status)) << TF_Message(status);
    TFE_Execute(matmul, &retvals[0], &num_retvals, status);
    CHECK_EQ(TF_OK, TF_GetCode(status)) << TF_Message(status);