imperative.cc 25.3 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
/* Copyright (c) 2018 PaddlePaddle Authors. All Rights Reserved.

Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at

http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License. */

#include "paddle/fluid/pybind/imperative.h"
16

17
#include <Python.h>
18 19 20 21
#include <pybind11/chrono.h>
#include <pybind11/complex.h>
#include <pybind11/functional.h>
#include <pybind11/stl.h>
22
#include <memory>
J
Jiabin Yang 已提交
23
#include <string>
24 25
#include <unordered_map>
#include <utility>
J
Jiabin Yang 已提交
26 27
#include <vector>
#include "paddle/fluid/imperative/backward_strategy.h"
28
#include "paddle/fluid/imperative/data_loader.h"
29
#include "paddle/fluid/imperative/layer.h"
J
Jiabin Yang 已提交
30
#include "paddle/fluid/imperative/nccl_context.h"
31
#include "paddle/fluid/imperative/profiler.h"
32
#include "paddle/fluid/imperative/tracer.h"
M
minqiyang 已提交
33
#include "paddle/fluid/imperative/type_defs.h"
34
#include "paddle/fluid/pybind/op_function.h"
35
#include "paddle/fluid/pybind/pybind_boost_headers.h"
L
Leo Chen 已提交
36
#include "paddle/fluid/pybind/tensor_py.h"
37

38 39 40
namespace paddle {
namespace pybind {

41 42
namespace py = ::pybind11;

43 44 45 46
class Layer : public imperative::Layer {
 public:
  using imperative::Layer::Layer;  // Inherit constructors

47 48 49 50
  std::vector<std::shared_ptr<imperative::VarBase>> Forward(
      const std::vector<std::shared_ptr<imperative::VarBase>> &inputs)
      override {
    PYBIND11_OVERLOAD(std::vector<std::shared_ptr<imperative::VarBase>>, Layer,
J
Jiabin Yang 已提交
51
                      Forward, inputs);  // NOLINT
52 53 54
  }
};

L
Leo Chen 已提交
55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77
static const platform::Place PyObjectToPlace(const py::object &place_obj) {
  if (py::isinstance<platform::CPUPlace>(place_obj)) {
    return place_obj.cast<platform::CPUPlace>();
  } else if (py::isinstance<platform::CUDAPlace>(place_obj)) {
    return place_obj.cast<platform::CUDAPlace>();
  } else if (py::isinstance<platform::CUDAPinnedPlace>(place_obj)) {
    return place_obj.cast<platform::CUDAPinnedPlace>();
  } else {
    PADDLE_THROW(platform::errors::InvalidArgument(
        "Place should be one of CPUPlace/CUDAPlace/CUDAPinnedPlace"));
  }
}

static void InitTensorForVarBase(imperative::VarBase *self,
                                 const py::array &array,
                                 const platform::Place place,
                                 bool persistable = false,
                                 bool zero_copy = false,
                                 std::string name = "") {
  if (name == "") {
    name = imperative::GetCurrentTracer()->GenerateUniqueName("generated_var");
  }
  new (self) imperative::VarBase(name);
78
  auto *tensor = self->MutableVar()->GetMutable<framework::LoDTensor>();
L
Leo Chen 已提交
79 80 81 82 83 84 85 86 87
  if (platform::is_cpu_place(place)) {
    SetTensorFromPyArray<platform::CPUPlace>(
        tensor, array, boost::get<platform::CPUPlace>(place), zero_copy);
  } else if (platform::is_gpu_place(place)) {
    SetTensorFromPyArray<platform::CUDAPlace>(
        tensor, array, boost::get<platform::CUDAPlace>(place), zero_copy);
  } else if (platform::is_cuda_pinned_place(place)) {
    SetTensorFromPyArray<platform::CUDAPinnedPlace>(
        tensor, array, boost::get<platform::CUDAPinnedPlace>(place), zero_copy);
88
  } else {
L
Leo Chen 已提交
89 90
    PADDLE_THROW(platform::errors::InvalidArgument(
        "Place should be one of CPUPlace/CUDAPlace/CUDAPinnedPlace"));
J
Jiabin Yang 已提交
91
  }
L
Leo Chen 已提交
92
  self->SetPersistable(persistable);
93 94 95 96 97 98 99 100 101
  self->SetType(framework::proto::VarType::LOD_TENSOR);
  self->SetDataType(tensor->type());
}

static void InitVarBaseFromNumpyWithKwargs(imperative::VarBase *self,
                                           const py::kwargs &kwargs) {
  PADDLE_ENFORCE_EQ(
      kwargs.contains("value"), true,
      platform::errors::InvalidArgument("Missing argument: value"));
L
Leo Chen 已提交
102 103 104 105 106 107 108 109 110 111 112 113 114

  auto persistable = kwargs.contains("persistable")
                         ? kwargs["persistable"].cast<bool>()
                         : false;
  auto array = kwargs.contains("value") ? kwargs["value"].cast<py::array>()
                                        : py::array();
  auto zero_copy =
      kwargs.contains("zero_copy") ? kwargs["zero_copy"].cast<bool>() : false;
  auto name = kwargs.contains("name") ? kwargs["name"].cast<std::string>() : "";
  auto default_place = imperative::GetCurrentTracer()->ExpectedPlace();
  auto place = kwargs.contains("place") ? PyObjectToPlace(kwargs["place"])
                                        : default_place;
  InitTensorForVarBase(self, array, place, persistable, zero_copy, name);
115
}
116

117 118 119
template <typename P>
static void InitVarBaseFromNumpyWithArg(imperative::VarBase *self,
                                        const py::array &array, const P &place,
L
Leo Chen 已提交
120 121 122 123 124 125 126 127
                                        bool persistable = false,
                                        bool zero_copy = false,
                                        std::string name = "") {
  // 0: self, 1: value, 2: place, 3: persistable, 4: zero_copy, 5: name
  if (name == "") {
    name = imperative::GetCurrentTracer()->GenerateUniqueName("generated_var");
  }
  new (self) imperative::VarBase(name);
128 129 130 131 132 133 134 135
  self->SetPersistable(persistable);
  auto *tensor = self->MutableVar()->GetMutable<framework::LoDTensor>();
  SetTensorFromPyArray<P>(tensor, array, place, zero_copy);
  self->SetType(framework::proto::VarType::LOD_TENSOR);
  self->SetDataType(tensor->type());
}

static void InitVarBaseFromNumpyWithArgDefault(imperative::VarBase *self,
L
Leo Chen 已提交
136 137 138
                                               const py::array &array) {
  auto place = imperative::GetCurrentTracer()->ExpectedPlace();
  InitTensorForVarBase(self, array, place);
139
}
140

141 142 143 144 145
static std::string GetTypeName(const imperative::VarBase &var) {
  if (var.Type() == framework::proto::VarType::RAW) {
    return "RAW";
  } else if (!var.Var().IsInitialized()) {
    return "nullptr";
146
  } else {
147
    return framework::ToTypeName(var.Var().Type());
148 149
  }
}
L
Leo Chen 已提交
150

151
using PyNameVarBaseMap = std::unordered_map<std::string, py::handle>;
152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173

template <typename T>
static T PyObjectCast(PyObject *obj) {
  try {
    return py::cast<T>(py::handle(obj));
  } catch (py::cast_error &) {
    PADDLE_THROW("Python object is not type of %s", typeid(T).name());
  }
}

// NOTE(zjl): py::handle is a very light wrapper of PyObject *.
// Unlike py::object, py::handle does not change reference count of PyObject *.
static std::vector<std::shared_ptr<imperative::VarBase>>
GetVarBaseListFromPyHandle(const py::handle &handle) {
  PyObject *py_obj = handle.ptr();  // get underlying PyObject
  // Python None is not nullptr in C++!
  if (!py_obj || py_obj == Py_None) {
    return {};
  }

  std::vector<std::shared_ptr<imperative::VarBase>> result;

174
  if (PyList_Check(py_obj)) {  // List of VarBase
175 176 177
    size_t len = PyList_GET_SIZE(py_obj);
    result.reserve(len);
    for (size_t i = 0; i < len; ++i) {
178 179 180
      PyObject *py_ivar = PyList_GET_ITEM(py_obj, i);
      PADDLE_ENFORCE_NOT_NULL(
          py_ivar, platform::errors::InvalidArgument("Python Object is NULL"));
181 182 183
      result.emplace_back(
          PyObjectCast<std::shared_ptr<imperative::VarBase>>(py_ivar));
    }
184
  } else if (PyTuple_Check(py_obj)) {  // Tuple of VarBase
185 186 187
    size_t len = PyTuple_GET_SIZE(py_obj);
    result.reserve(len);
    for (size_t i = 0; i < len; ++i) {
188 189 190
      PyObject *py_ivar = PyTuple_GET_ITEM(py_obj, i);
      PADDLE_ENFORCE_NOT_NULL(
          py_ivar, platform::errors::InvalidArgument("Python Object is NULL"));
191 192 193
      result.emplace_back(
          PyObjectCast<std::shared_ptr<imperative::VarBase>>(py_ivar));
    }
194 195 196
  } else {  // VarBase
    result.emplace_back(
        PyObjectCast<std::shared_ptr<imperative::VarBase>>(py_obj));
197 198 199 200 201
  }

  return result;
}

J
Jiabin Yang 已提交
202 203 204
static imperative::NameVarBaseMap ConvertToNameVarBaseMap(
    const PyNameVarBaseMap &map) {
  imperative::NameVarBaseMap result;
205 206 207 208 209 210
  for (auto &pair : map) {
    auto var_vec = GetVarBaseListFromPyHandle(pair.second);
    if (!var_vec.empty()) {
      result.emplace(pair.first, std::move(var_vec));
    }
  }
J
Jiabin Yang 已提交
211 212 213

  PADDLE_ENFORCE_EQ(PyErr_Occurred() == nullptr, true,
                    py::str(py::handle(PyErr_Occurred())));
214 215 216
  return result;
}

217
// Bind Methods
J
Jiabin Yang 已提交
218
void BindImperative(py::module *m_ptr) {
219 220
  auto &m = *m_ptr;

221 222
  BindOpFunctions(&m);

223
  py::class_<imperative::detail::BackwardStrategy> backward_strategy(
224 225
      m, "BackwardStrategy", R"DOC(

J
Jiabin Yang 已提交
226
    BackwardStrategy is a descriptor of how to run the backward process.
227

J
Jiabin Yang 已提交
228
    **Note**:
T
tianshuo78520a 已提交
229
        **This API is only available in** `Dygraph <../../user_guides/howto/dygraph/DyGraph.html>`_ **Mode**
230

J
Jiabin Yang 已提交
231 232
    Attribute:
        **sort_sum_gradient**:
233

J
Jiabin Yang 已提交
234
        If framework will sum the gradient by the reverse order of trace. eg. x_var ( :ref:`api_guide_Variable` ) will be the input of multiple OP such as :ref:`api_fluid_layers_scale` , this attr will decide if framework will sum gradient of `x_var` by the reverse order.
L
lujun 已提交
235

J
Jiabin Yang 已提交
236
        By Default: False
L
lujun 已提交
237

J
Jiabin Yang 已提交
238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255
        Examples:
            .. code-block:: python

                import numpy as np
                import paddle.fluid as fluid

                x = np.ones([2, 2], np.float32)
                with fluid.dygraph.guard():
                    x_var = fluid.dygraph.to_variable(x)
                    sums_inputs = []
                    # x_var will be multi-scales' input here
                    for _ in range(10):
                        sums_inputs.append(fluid.layers.scale(x_var))
                    ret2 = fluid.layers.sums(sums_inputs)
                    loss2 = fluid.layers.reduce_sum(ret2)
                    backward_strategy = fluid.dygraph.BackwardStrategy()
                    backward_strategy.sort_sum_gradient = True
                    loss2.backward(backward_strategy)
256
      )DOC");
257 258 259 260 261 262 263 264 265 266 267 268 269 270 271
  backward_strategy.def(py::init())
      .def_property("sort_sum_gradient",
                    [](const imperative::detail::BackwardStrategy &self) {
                      return self.sorted_sum_gradient_;
                    },
                    [](imperative::detail::BackwardStrategy &self,
                       bool sorted_sum_gradient) {
                      self.sorted_sum_gradient_ = sorted_sum_gradient;
                    });

  m.def("start_imperative_gperf_profiler",
        []() { imperative::StartProfile(); });

  m.def("stop_imperative_gperf_profiler", []() { imperative::StopProfile(); });

Z
Zeng Jinle 已提交
272 273 274
  m.def("_is_dygraph_debug_enabled",
        []() { return imperative::IsDebugEnabled(); });
  m.def("_dygraph_debug_level", []() { return imperative::GetDebugLevel(); });
275 276 277 278
  m.def("_switch_tracer",
        [](const std::shared_ptr<imperative::Tracer> &tracer) {
          imperative::SetCurrentTracer(tracer);
        });
Z
Zeng Jinle 已提交
279

280 281 282 283 284 285 286 287 288 289 290 291 292
#ifndef _WIN32
  // Dygraph DataLoader signal handler
  m.def("_set_process_pid", [](int64_t key, pid_t pid) {
    imperative::SetLoadProcessPID(key, pid);
  });
  m.def("_erase_process_pid",
        [](int64_t key) { imperative::EraseLoadProcessPID(key); });
  m.def("_set_process_signal_handler",
        []() { imperative::SetLoadProcessSignalHandler(); });
  m.def("_throw_error_if_process_failed",
        []() { imperative::ThrowErrorIfLoadProcessFailed(); });
#endif

293
  py::class_<imperative::VarBase, std::shared_ptr<imperative::VarBase>>(
J
Jiabin Yang 已提交
294 295
      m, "VarBase",
      R"DOC()DOC")
Z
Zeng Jinle 已提交
296
      .def_static("_alive_vars", &imperative::VarBase::AliveVarNames)
J
Jiabin Yang 已提交
297
      .def("__init__",
298 299 300 301 302 303 304 305 306 307 308
           [](imperative::VarBase &self, framework::proto::VarType::Type dtype,
              const std::vector<int> &dims, const py::handle &name,
              framework::proto::VarType::Type type, bool persistable) {
             std::string act_name = "";
             if (!name.ptr() || name.ptr() == Py_None) {
               act_name = imperative::GetCurrentTracer()->GenerateUniqueName(
                   "generated_var");
             } else {
               act_name = name.cast<std::string>();
             }
             new (&self) imperative::VarBase(act_name);
J
Jiabin Yang 已提交
309 310 311 312 313 314 315 316 317
             self.SetPersistable(persistable);
             self.SetType(type);
             self.SetDataType(dtype);
             if (type == framework::proto::VarType::LOD_TENSOR) {
               auto *tensor =
                   self.MutableVar()->GetMutable<framework::LoDTensor>();
               tensor->Resize(framework::make_ddim(dims));
             }
           })
318 319
      .def("__init__", &InitVarBaseFromNumpyWithArg<platform::CPUPlace>,
           py::arg("value"), py::arg("place"), py::arg("persistable") = false,
L
Leo Chen 已提交
320
           py::arg("zero_copy") = false, py::arg("name") = "")
321 322
      .def("__init__", &InitVarBaseFromNumpyWithArg<platform::CUDAPlace>,
           py::arg("value"), py::arg("place"), py::arg("persistable") = false,
L
Leo Chen 已提交
323
           py::arg("zero_copy") = false, py::arg("name") = "")
324 325
      .def("__init__", &InitVarBaseFromNumpyWithArg<platform::CUDAPinnedPlace>,
           py::arg("value"), py::arg("place"), py::arg("persistable") = false,
L
Leo Chen 已提交
326 327
           py::arg("zero_copy") = false, py::arg("name") = "")
      .def("__init__", &InitVarBaseFromNumpyWithArgDefault, py::arg("value"))
328 329 330 331 332 333 334 335 336 337 338 339 340 341
      .def("__init__", &InitVarBaseFromNumpyWithKwargs)
      .def("numpy",
           [](imperative::VarBase &self) -> py::array {
             const auto &tensor =
                 self.MutableVar()->Get<framework::LoDTensor>();
             PADDLE_ENFORCE_EQ(
                 tensor.IsInitialized(), true,
                 platform::errors::InvalidArgument(
                     "%s is Empty, Please check if it has no data in",
                     self.Name()));
             return TensorToPyArray(tensor, true);
           },
           R"DOC(
        **Notes**:
T
tianshuo78520a 已提交
342
            **This API is ONLY available in Dygraph mode**
343 344 345 346 347 348 349 350 351 352 353 354 355 356

        Returns a numpy array shows the value of current :ref:`api_guide_Variable_en`

        Returns:
            ndarray: The numpy value of current Variable.

        Returns type:
            ndarray: dtype is same as current Variable

        Examples:
            .. code-block:: python

                import paddle.fluid as fluid
                from paddle.fluid.dygraph.base import to_variable
357
                from paddle.fluid.dygraph import Linear
358 359 360 361
                import numpy as np

                data = np.random.uniform(-1, 1, [30, 10, 32]).astype('float32')
                with fluid.dygraph.guard():
362
                    linear = Linear(32, 64)
363
                    data = to_variable(data)
364
                    x = linear(data)
365 366 367 368 369 370 371 372 373 374 375 376 377
                    print(x.numpy())

       )DOC")
      .def("detach",
           [](const imperative::VarBase &self) {
             const auto &tensor = self.Var().Get<framework::LoDTensor>();
             PADDLE_ENFORCE_EQ(tensor.IsInitialized(), true,
                               platform::errors::InvalidArgument(
                                   "%s has not been initialized", self.Name()));
             return self.NewVarBase(tensor.place(), false);
           },
           py::return_value_policy::copy, R"DOC(
        **Notes**:
T
tianshuo78520a 已提交
378
            **This API is ONLY available in Dygraph mode**
379 380 381 382 383 384 385 386 387 388 389 390

        Returns a new Variable, detached from the current graph.

        Returns:
             ( :ref:`api_guide_Variable_en` | dtype is same as current Variable): The detached Variable.


        Examples:
            .. code-block:: python

                import paddle.fluid as fluid
                from paddle.fluid.dygraph.base import to_variable
391
                from paddle.fluid.dygraph import Linear
392 393 394 395
                import numpy as np

                data = np.random.uniform(-1, 1, [30, 10, 32]).astype('float32')
                with fluid.dygraph.guard():
396
                    linear = Linear(32, 64)
397
                    data = to_variable(data)
398
                    x = linear(data)
399 400 401 402 403 404
                    y = x.detach()

       )DOC")
      .def("clear_gradient", &imperative::VarBase::ClearGradient, R"DOC(

        **Notes**:
T
tianshuo78520a 已提交
405
        **1. This API is ONLY available in Dygraph mode**
406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434

        **2. Use it only Variable has gradient, normally we use this for Parameters since other temporal Variable will be deleted by Python's GC**

        Clear  (set to ``0`` ) the Gradient of Current Variable

        Returns:  None

        Examples:
             .. code-block:: python

                import paddle.fluid as fluid
                import numpy as np

                x = np.ones([2, 2], np.float32)
                with fluid.dygraph.guard():
                    inputs2 = []
                    for _ in range(10):
                         tmp = fluid.dygraph.base.to_variable(x)
                         tmp.stop_gradient=False
                         inputs2.append(tmp)
                    ret2 = fluid.layers.sums(inputs2)
                    loss2 = fluid.layers.reduce_sum(ret2)
                    backward_strategy = fluid.dygraph.BackwardStrategy()
                    backward_strategy.sort_sum_gradient = True
                    loss2.backward(backward_strategy)
                    print(loss2.gradient())
                    loss2.clear_gradient()
                    print("After clear {}".format(loss2.gradient()))
      )DOC")
L
Leo Chen 已提交
435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454
      .def("_run_backward",
           [](imperative::VarBase &self,
              const imperative::detail::BackwardStrategy &bckst,
              const imperative::Tracer &tracer) {
             // TODO(jiabin): when we impl more backward execution we can select
             // them

             imperative::Engine *engine = tracer.GetDefaultEngine();
             VLOG(3) << "Start backward";
             engine->Init(&self, bckst);
             engine->Execute();
             VLOG(3) << "Finish backward";
           },
           py::call_guard<py::gil_scoped_release>())
      .def("_grad_name", &imperative::VarBase::GradVarName)
      .def("_grad_value",
           [](imperative::VarBase &self) {
             return self.MutableGradVar()->Get<framework::LoDTensor>();
           },
           py::return_value_policy::reference)
455
      .def("_grad_ivar",
J
Jiabin Yang 已提交
456 457
           [](const imperative::VarBase &self) {
             auto &grad_var = self.GradVarBase();
458 459 460 461 462 463 464 465 466 467 468
             if (grad_var && grad_var->Var().IsInitialized()) {
               auto *tensor =
                   grad_var->MutableVar()->IsType<framework::LoDTensor>()
                       ? grad_var->MutableVar()
                             ->GetMutable<framework::LoDTensor>()
                       : grad_var->MutableVar()
                             ->GetMutable<framework::SelectedRows>()
                             ->mutable_value();
               if (tensor->IsInitialized()) {
                 return grad_var;
               }
J
Jiabin Yang 已提交
469
             }
470
             return std::shared_ptr<imperative::VarBase>(nullptr);
J
Jiabin Yang 已提交
471 472
           },
           py::return_value_policy::copy)
473 474
      .def("_copy_to",
           [](const imperative::VarBase &self, const platform::CPUPlace &place,
J
Jiabin Yang 已提交
475 476
              bool blocking) { return self.NewVarBase(place, blocking); },
           py::return_value_policy::copy)
477 478
      .def("_copy_to",
           [](const imperative::VarBase &self, const platform::CUDAPlace &place,
J
Jiabin Yang 已提交
479 480 481
              bool blocking) { return self.NewVarBase(place, blocking); },
           py::return_value_policy::copy)
      .def("value", [](imperative::VarBase &self) { return self.MutableVar(); },
482 483 484
           py::return_value_policy::reference)
      .def_property("name", &imperative::VarBase::Name,
                    &imperative::VarBase::SetName)
L
Leo Chen 已提交
485 486 487 488 489
      .def_property("stop_gradient",
                    &imperative::VarBase::OverridedStopGradient,
                    &imperative::VarBase::SetOverridedStopGradient)
      .def_property("persistable", &imperative::VarBase::Persistable,
                    &imperative::VarBase::SetPersistable)
J
Jiabin Yang 已提交
490 491 492 493
      .def_property_readonly(
          "shape",
          [](imperative::VarBase &self) {
            if (self.Var().IsType<framework::LoDTensor>()) {
494
              return framework::vectorize<int>(
J
Jiabin Yang 已提交
495
                  self.Var().Get<framework::LoDTensor>().dims());
496 497 498
            } else if (self.Var().IsType<framework::SelectedRows>()) {
              return framework::vectorize<int>(
                  self.Var().Get<framework::SelectedRows>().value().dims());
J
Jiabin Yang 已提交
499 500 501 502 503 504 505
            } else {
              VLOG(2) << "It is meaningless to get shape of variable type "
                      << GetTypeName(self);
              return std::vector<int>();
            }
          })
      .def_property_readonly("type", &imperative::VarBase::Type)
L
Leo Chen 已提交
506
      .def_property_readonly("dtype", &imperative::VarBase::DataType);
507 508 509

  py::class_<imperative::Layer, Layer /* <--- trampoline*/> layer(m, "Layer");
  layer.def(py::init<>())
510 511 512 513 514
      .def("forward",
           [](imperative::Layer &self,
              const std::vector<std::shared_ptr<imperative::VarBase>> &inputs) {
             return self.Forward(inputs);
           });
515

516 517 518 519 520
  py::class_<imperative::jit::ProgramDescTracer>(m, "ProgramDescTracer", "")
      .def("create_program_desc",
           &imperative::jit::ProgramDescTracer::CreateProgramDesc)
      .def("reset", &imperative::jit::ProgramDescTracer::Reset);

521 522 523
  py::class_<imperative::Tracer, std::shared_ptr<imperative::Tracer>>(
      m, "Tracer",
      R"DOC()DOC")
524
      .def("__init__",
J
Jiabin Yang 已提交
525
           [](imperative::Tracer &self) { new (&self) imperative::Tracer(); })
526 527 528
      .def_property("_enable_program_desc_tracing",
                    &imperative::Tracer::IsProgramDescTracingEnabled,
                    &imperative::Tracer::SetEnableProgramDescTracing)
529 530 531 532 533 534 535 536 537 538
      .def_property("_train_mode", &imperative::Tracer::NoGrad,
                    &imperative::Tracer::SetNoGrad)
      .def_property(
          "_expected_place",
          [](const imperative::Tracer &self) -> py::object {
            return py::cast(self.ExpectedPlace());
          },
          [](imperative::Tracer &self, const py::object &obj) {
            if (py::isinstance<platform::CUDAPlace>(obj)) {
              auto p = obj.cast<platform::CUDAPlace *>();
L
Leo Chen 已提交
539
              self.SetExpectedPlace(*p);
540 541
            } else if (py::isinstance<platform::CPUPlace>(obj)) {
              auto p = obj.cast<platform::CPUPlace *>();
L
Leo Chen 已提交
542
              self.SetExpectedPlace(*p);
543 544
            } else if (py::isinstance<platform::CUDAPinnedPlace>(obj)) {
              auto p = obj.cast<platform::CUDAPinnedPlace *>();
L
Leo Chen 已提交
545
              self.SetExpectedPlace(*p);
546
            } else {
L
Leo Chen 已提交
547
              PADDLE_THROW(platform::errors::InvalidArgument(
548
                  "Incompatible Place Type: supports CUDAPlace, CPUPlace, "
L
Leo Chen 已提交
549 550
                  "and CUDAPinnedPlace, "
                  "but got Unknown Type!"));
551 552
            }
          })
553 554 555
      .def("_get_program_desc_tracer",
           &imperative::Tracer::GetProgramDescTracer,
           py::return_value_policy::reference)
M
minqiyang 已提交
556
      .def("trace",
J
Jiabin Yang 已提交
557 558 559 560 561 562
           [](imperative::Tracer &self, const std::string &type,
              const PyNameVarBaseMap &ins, const PyNameVarBaseMap &outs,
              framework::AttributeMap attrs, const platform::CUDAPlace &place,
              bool trace_backward) {
             auto ins_map = ConvertToNameVarBaseMap(ins);
             auto outs_map = ConvertToNameVarBaseMap(outs);
563 564
             {
               py::gil_scoped_release release;
J
Jiabin Yang 已提交
565 566
               self.TraceOp(type, std::move(ins_map), std::move(outs_map),
                            std::move(attrs), place, trace_backward);
567
             }
M
minqiyang 已提交
568
           })
J
Jiabin Yang 已提交
569 570 571 572 573 574 575 576 577 578 579 580 581
      .def("trace",
           [](imperative::Tracer &self, const std::string &type,
              const PyNameVarBaseMap &ins, const PyNameVarBaseMap &outs,
              framework::AttributeMap attrs, const platform::CPUPlace &place,
              bool trace_backward) {
             auto ins_map = ConvertToNameVarBaseMap(ins);
             auto outs_map = ConvertToNameVarBaseMap(outs);
             {
               py::gil_scoped_release release;
               self.TraceOp(type, std::move(ins_map), std::move(outs_map),
                            std::move(attrs), place, trace_backward);
             }
           });
582 583

  // define parallel context
584 585 586
  py::class_<imperative::ParallelStrategy> parallel_strategy(
      m, "ParallelStrategy", "");
  parallel_strategy.def(py::init())
587 588
      .def_property(
          "nranks",
589 590
          [](const imperative::ParallelStrategy &self) { return self.nranks_; },
          [](imperative::ParallelStrategy &self, int nranks) {
591 592 593
            self.nranks_ = nranks;
          })
      .def_property("local_rank",
594
                    [](const imperative::ParallelStrategy &self) {
595 596
                      return self.local_rank_;
                    },
597
                    [](imperative::ParallelStrategy &self, int local_rank) {
598 599 600 601
                      self.local_rank_ = local_rank;
                    })
      .def_property(
          "trainer_endpoints",
602
          [](const imperative::ParallelStrategy &self) {
603 604
            return self.trainer_endpoints_;
          },
605
          [](imperative::ParallelStrategy &self, std::vector<std::string> eps) {
606 607 608
            self.trainer_endpoints_ = eps;
          })
      .def_property("current_endpoint",
609
                    [](const imperative::ParallelStrategy &self) {
610 611
                      return self.current_endpoint_;
                    },
612 613
                    [](imperative::ParallelStrategy &self,
                       const std::string &ep) { self.current_endpoint_ = ep; });
614
#if defined(PADDLE_WITH_NCCL)
615 616
  py::class_<imperative::NCCLParallelContext> nccl_ctx(m,
                                                       "NCCLParallelContext");
617 618

  nccl_ctx
619 620 621
      .def(py::init<const imperative::ParallelStrategy &,
                    const platform::CUDAPlace &>())
      .def("init", [](imperative::NCCLParallelContext &self) { self.Init(); });
622
#endif
623 624 625 626
}

}  // namespace pybind
}  // namespace paddle