imperative.cc 49.3 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

23
#include <memory>
24
#include <set>
J
Jiabin Yang 已提交
25
#include <string>
26
#include <unordered_map>
27
#include <unordered_set>
28
#include <utility>
J
Jiabin Yang 已提交
29
#include <vector>
30

31
#include "paddle/fluid/imperative/all_reduce.h"
32
#include "paddle/fluid/imperative/amp_auto_cast.h"
33
#include "paddle/fluid/imperative/basic_engine.h"
34
#include "paddle/fluid/imperative/data_loader.h"
35
#include "paddle/fluid/imperative/layer.h"
J
Jiabin Yang 已提交
36
#include "paddle/fluid/imperative/nccl_context.h"
37
#include "paddle/fluid/imperative/partial_grad_engine.h"
38
#include "paddle/fluid/imperative/profiler.h"
39
#include "paddle/fluid/imperative/tracer.h"
M
minqiyang 已提交
40
#include "paddle/fluid/imperative/type_defs.h"
41
#include "paddle/fluid/memory/allocation/mmap_allocator.h"
42
#include "paddle/fluid/pybind/op_function.h"
43
#include "paddle/fluid/pybind/pybind_boost_headers.h"
L
Leo Chen 已提交
44
#include "paddle/fluid/pybind/tensor_py.h"
45

46 47 48
namespace paddle {
namespace pybind {

49 50
namespace py = ::pybind11;

51 52 53 54
class Layer : public imperative::Layer {
 public:
  using imperative::Layer::Layer;  // Inherit constructors

55 56 57 58
  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
59
                      Forward, inputs);  // NOLINT
60 61 62
  }
};

L
Leo Chen 已提交
63 64 65 66 67
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>();
68 69
  } else if (py::isinstance<platform::XPUPlace>(place_obj)) {
    return place_obj.cast<platform::XPUPlace>();
L
Leo Chen 已提交
70 71
  } else if (py::isinstance<platform::CUDAPinnedPlace>(place_obj)) {
    return place_obj.cast<platform::CUDAPinnedPlace>();
72 73
  } else if (py::isinstance<platform::Place>(place_obj)) {
    return place_obj.cast<platform::Place>();
L
Leo Chen 已提交
74 75
  } else {
    PADDLE_THROW(platform::errors::InvalidArgument(
76 77
        "Place should be one of "
        "Place/CPUPlace/XPUPlace/CUDAPlace/CUDAPinnedPlace"));
L
Leo Chen 已提交
78 79 80 81 82 83 84
  }
}

static void InitTensorForVarBase(imperative::VarBase *self,
                                 const py::array &array,
                                 const platform::Place place,
                                 bool persistable = false,
85 86
                                 bool zero_copy = false, std::string name = "",
                                 int stop_gradient = -1) {
L
Leo Chen 已提交
87
  if (name == "") {
88 89
    name =
        imperative::GetCurrentTracer()->GenerateUniqueName("generated_tensor");
L
Leo Chen 已提交
90
  }
91 92 93
  VLOG(5) << "Init Tensor as: / name: " << name
          << " / persistable: " << persistable << " / zero_copy: " << zero_copy
          << " / stop_gradient: " << stop_gradient;
L
Leo Chen 已提交
94
  new (self) imperative::VarBase(name);
95
  auto *tensor = self->MutableVar()->GetMutable<framework::LoDTensor>();
L
Leo Chen 已提交
96 97
  if (platform::is_cpu_place(place)) {
    SetTensorFromPyArray<platform::CPUPlace>(
98
        tensor, array, BOOST_GET_CONST(platform::CPUPlace, place), zero_copy);
99 100 101
  } else if (platform::is_xpu_place(place)) {
    SetTensorFromPyArray<platform::XPUPlace>(
        tensor, array, BOOST_GET_CONST(platform::XPUPlace, place), zero_copy);
L
Leo Chen 已提交
102 103
  } else if (platform::is_gpu_place(place)) {
    SetTensorFromPyArray<platform::CUDAPlace>(
104
        tensor, array, BOOST_GET_CONST(platform::CUDAPlace, place), zero_copy);
L
Leo Chen 已提交
105 106
  } else if (platform::is_cuda_pinned_place(place)) {
    SetTensorFromPyArray<platform::CUDAPinnedPlace>(
107 108
        tensor, array, BOOST_GET_CONST(platform::CUDAPinnedPlace, place),
        zero_copy);
109
  } else {
L
Leo Chen 已提交
110
    PADDLE_THROW(platform::errors::InvalidArgument(
111
        "Place should be one of CPUPlace/XPUPlace/CUDAPlace/CUDAPinnedPlace"));
J
Jiabin Yang 已提交
112
  }
113 114 115
  if (stop_gradient != -1) {
    self->SetOverridedStopGradient(stop_gradient);
  }
L
Leo Chen 已提交
116
  self->SetPersistable(persistable);
117 118 119 120 121 122
  self->SetType(framework::proto::VarType::LOD_TENSOR);
  self->SetDataType(tensor->type());
}

static void InitVarBaseFromNumpyWithKwargs(imperative::VarBase *self,
                                           const py::kwargs &kwargs) {
123
  VLOG(4) << "Init VarBase from kwargs: ";
124 125
  PADDLE_ENFORCE_EQ(
      kwargs.contains("value"), true,
126 127
      platform::errors::NotFound(
          "The kwargs used to create Varbase misses argument: value"));
L
Leo Chen 已提交
128 129 130 131 132 133 134 135
  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>() : "";
136 137 138
  auto stop_gradient = kwargs.contains("stop_gradient")
                           ? kwargs["stop_gradient"].cast<int>()
                           : -1;
L
Leo Chen 已提交
139 140 141
  auto default_place = imperative::GetCurrentTracer()->ExpectedPlace();
  auto place = kwargs.contains("place") ? PyObjectToPlace(kwargs["place"])
                                        : default_place;
142 143
  InitTensorForVarBase(self, array, place, persistable, zero_copy, name,
                       stop_gradient);
144
}
145

146 147 148
template <typename P>
static void InitVarBaseFromNumpyWithArg(imperative::VarBase *self,
                                        const py::array &array, const P &place,
L
Leo Chen 已提交
149 150
                                        bool persistable = false,
                                        bool zero_copy = false,
151 152 153 154 155
                                        std::string name = "",
                                        int stop_gradient = -1) {
  VLOG(4) << "Init VarBase from Arg: ";
  // 0: self, 1: value, 2: place, 3: persistable, 4: zero_copy, 5: name , 6:
  // stop_gradient
L
Leo Chen 已提交
156
  if (name == "") {
157 158
    name =
        imperative::GetCurrentTracer()->GenerateUniqueName("generated_tensor");
L
Leo Chen 已提交
159
  }
160 161 162
  VLOG(5) << "Init Tensor as: / name: " << name
          << " / persistable: " << persistable << " / zero_copy: " << zero_copy
          << " / stop_gradient: " << stop_gradient;
L
Leo Chen 已提交
163
  new (self) imperative::VarBase(name);
164 165
  self->SetPersistable(persistable);
  auto *tensor = self->MutableVar()->GetMutable<framework::LoDTensor>();
166 167 168
  if (stop_gradient != -1) {
    self->SetOverridedStopGradient(stop_gradient);
  }
169 170 171 172 173 174
  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
175
                                               const py::array &array) {
176
  VLOG(4) << "Init VarBase from numpy: ";
L
Leo Chen 已提交
177 178
  auto place = imperative::GetCurrentTracer()->ExpectedPlace();
  InitTensorForVarBase(self, array, place);
179
}
180

181 182 183 184 185
static void InitVarBaseFromTensorWithArgDefault(
    imperative::VarBase *self, const framework::LoDTensor &tensor) {
  VLOG(4) << "Init VarBase";
  auto place = imperative::GetCurrentTracer()->ExpectedPlace();
  new (self) imperative::VarBase(
186
      imperative::GetCurrentTracer()->GenerateUniqueName("generated_tensor"));
187 188 189 190 191 192 193 194 195 196 197 198 199 200
  self->SetPersistable(false);
  self->SetType(framework::proto::VarType::LOD_TENSOR);
  self->SetDataType(tensor.type());
  auto *new_tensor = self->MutableVar()->GetMutable<framework::LoDTensor>();
  // Same place,share data directly
  if (place == tensor.place()) {
    new_tensor->ShareDataWith(tensor);
    VLOG(4) << "Same place, do ShareDataWith";
  } else {
    framework::TensorCopy(tensor, place, new_tensor);
    VLOG(4) << "Different place, do TensorCopy";
  }
}

201 202 203 204 205
static std::string GetTypeName(const imperative::VarBase &var) {
  if (var.Type() == framework::proto::VarType::RAW) {
    return "RAW";
  } else if (!var.Var().IsInitialized()) {
    return "nullptr";
206
  } else {
207
    return framework::ToTypeName(var.Var().Type());
208 209
  }
}
L
Leo Chen 已提交
210

211
using PyNameVarBaseMap = std::unordered_map<std::string, py::handle>;
212 213 214 215 216 217

template <typename T>
static T PyObjectCast(PyObject *obj) {
  try {
    return py::cast<T>(py::handle(obj));
  } catch (py::cast_error &) {
218 219
    PADDLE_THROW(platform::errors::InvalidArgument(
        "Python object is not type of %s", typeid(T).name()));
220 221 222 223 224 225 226 227 228 229 230 231 232 233 234
  }
}

// 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;

235
  if (PyList_Check(py_obj)) {  // List of VarBase
236 237 238
    size_t len = PyList_GET_SIZE(py_obj);
    result.reserve(len);
    for (size_t i = 0; i < len; ++i) {
239 240 241
      PyObject *py_ivar = PyList_GET_ITEM(py_obj, i);
      PADDLE_ENFORCE_NOT_NULL(
          py_ivar, platform::errors::InvalidArgument("Python Object is NULL"));
242 243 244
      result.emplace_back(
          PyObjectCast<std::shared_ptr<imperative::VarBase>>(py_ivar));
    }
245
  } else if (PyTuple_Check(py_obj)) {  // Tuple of VarBase
246 247 248
    size_t len = PyTuple_GET_SIZE(py_obj);
    result.reserve(len);
    for (size_t i = 0; i < len; ++i) {
249 250 251
      PyObject *py_ivar = PyTuple_GET_ITEM(py_obj, i);
      PADDLE_ENFORCE_NOT_NULL(
          py_ivar, platform::errors::InvalidArgument("Python Object is NULL"));
252 253 254
      result.emplace_back(
          PyObjectCast<std::shared_ptr<imperative::VarBase>>(py_ivar));
    }
255 256 257
  } else {  // VarBase
    result.emplace_back(
        PyObjectCast<std::shared_ptr<imperative::VarBase>>(py_obj));
258 259 260 261 262
  }

  return result;
}

J
Jiabin Yang 已提交
263 264 265
static imperative::NameVarBaseMap ConvertToNameVarBaseMap(
    const PyNameVarBaseMap &map) {
  imperative::NameVarBaseMap result;
266 267 268 269 270 271
  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
272

273 274 275
  PADDLE_ENFORCE_EQ(
      PyErr_Occurred(), nullptr,
      platform::errors::InvalidArgument(py::str(py::handle(PyErr_Occurred()))));
276 277 278
  return result;
}

279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343
static bool PyCheckInteger(PyObject *obj) {
#if PY_VERSION_HEX < 0x03000000
  return (PyLong_Check(obj) || PyInt_Check(obj)) && !PyBool_Check(obj);
#else
  return PyLong_Check(obj) && !PyBool_Check(obj);
#endif
}

// NOTE(zhiqiu): Revised version of PySlice_GetIndices. From:
// https://github.com/python/cpython/blob/8d21aa21f2cbc6d50aab3f420bb23be1d081dac4/Objects/sliceobject.c#L103
// Original PySlice_GetIndices return wrong result when
// slice_item contains long int, such as arr[:180L].
// NOT sure why this happens !!!
// Besides, PySlice_GetIndices cannot raise error when float in slice item.
// So, I make a revised version of PySlice_GetIndices, named to
// _PySlice_GetIndices. Try to use _PySlice_Unpack which is more robust than
// PySlice_GetIndices in the future.
static int _PySlice_GetIndices(PySliceObject *r, Py_ssize_t length,
                               Py_ssize_t *start, Py_ssize_t *stop,
                               Py_ssize_t *step) {
  /* XXX support long ints */
  if (r->step == Py_None) {
    *step = 1;
  } else {
    if (PyCheckInteger(r->step)) {
      *step = PyLong_AsLong(r->step);
    } else {
      PADDLE_THROW(platform::errors::InvalidArgument(
          "Currently, VarBase.__getitem__() only allows None or integers in "
          "slice item, but received %s.",
          std::string(Py_TYPE(r->step)->tp_name)));
    }
  }
  if (r->start == Py_None) {
    *start = *step < 0 ? length - 1 : 0;
  } else {
    if (PyCheckInteger(r->start)) {
      *start = PyLong_AsLong(r->start);
    } else {
      PADDLE_THROW(platform::errors::InvalidArgument(
          "Currently, VarBase.__getitem__() only allows None or integers in "
          "slice item, but received %s.",
          std::string(Py_TYPE(r->start)->tp_name)));
    }
    if (*start < 0) *start += length;
  }
  if (r->stop == Py_None) {
    *stop = *step < 0 ? -1 : length;
  } else {
    if (PyCheckInteger(r->stop)) {
      *stop = PyLong_AsLong(r->stop);
    } else {
      PADDLE_THROW(platform::errors::InvalidArgument(
          "Currently, VarBase.__getitem__() only allows None or integers in "
          "slice item, but received %s.",
          std::string(Py_TYPE(r->stop)->tp_name)));
    }
    if (*stop < 0) *stop += length;
  }
  if (*stop > length) return -1;
  if (*start >= length) return -1;
  if (*step == 0) return -1;
  return 0;
}

S
songyouwei 已提交
344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367
static void ParseIndexingSlice(framework::LoDTensor *tensor, PyObject *_index,
                               std::vector<int> *slice_axes,
                               std::vector<int> *slice_starts,
                               std::vector<int> *slice_ends,
                               std::vector<int> *slice_strides,
                               std::vector<int> *decrease_axis,
                               std::vector<int> *infer_flags) {
  // We allow indexing by Integers, Slices, and tuples of those
  // types.
  // Ellipsis and None are not supported yet.
  // wrap to tuple
  PyObject *index = !PyTuple_Check(_index) ? PyTuple_Pack(1, _index) : _index;
  PADDLE_ENFORCE_EQ(
      tensor->IsInitialized(), true,
      platform::errors::InvalidArgument("tensor has not been initialized"));
  const auto &shape = tensor->dims();
  const int rank = shape.size();
  const int size = PyTuple_GET_SIZE(index);
  PADDLE_ENFORCE_EQ(
      size <= rank, true,
      platform::errors::InvalidArgument(
          "too many indices (%d) for tensor of dimension %d", size, rank));
  for (int dim = 0; dim < size; ++dim) {
    PyObject *slice_item = PyTuple_GetItem(index, dim);
368 369 370 371 372 373 374
    PADDLE_ENFORCE_EQ(PyCheckInteger(slice_item) || PySlice_Check(slice_item),
                      true,
                      platform::errors::InvalidArgument(
                          "Currently, VarBase.__getitem__() only allows "
                          "indexing by Integers, Slices, and tuples of "
                          "these types, but received %s in %dth slice item",
                          std::string(Py_TYPE(slice_item)->tp_name), dim + 1));
S
songyouwei 已提交
375 376
    infer_flags->push_back(1);
    int dim_len = shape[dim];
377 378
    if (PyCheckInteger(slice_item)) {
      // integer, PyLong_AsLong supports both int and long
S
songyouwei 已提交
379
      int start = static_cast<int>(PyLong_AsLong(slice_item));
H
hong 已提交
380
      auto s_t = start;
S
songyouwei 已提交
381
      start = start < 0 ? start + dim_len : start;
H
hong 已提交
382 383 384 385 386 387 388 389 390 391
      if (start >= dim_len) {
        std::string str_error_message =
            "The starting index " + std::to_string(s_t) +
            " of slice is out of bounds in tensor " + std::to_string(dim) +
            "-th axis, it shound be in the range of [" +
            std::to_string(-dim_len) + ", " + std::to_string(dim_len) + ")";
        // py::index_error is corresponding to IndexError in Python
        // Used to indicate out of bounds access in __getitem__, __setitem__
        throw py::index_error(str_error_message);
      }
S
songyouwei 已提交
392 393 394 395 396 397
      slice_axes->push_back(dim);
      slice_starts->push_back(start);
      slice_ends->push_back(start + 1);
      slice_strides->push_back(1);
      decrease_axis->push_back(dim);
    } else {
398
      // slice item
S
songyouwei 已提交
399
      Py_ssize_t start, end, step;
400 401 402
      PySliceObject *p = reinterpret_cast<PySliceObject *>(slice_item);
      _PySlice_GetIndices(p, dim_len, &start, &end, &step);

S
songyouwei 已提交
403
      // :: or : or 0:dim_len:1
404 405 406
      if (start == 0 && end == dim_len && step == 1) {
        continue;
      }
S
songyouwei 已提交
407 408 409 410 411 412 413 414 415
      slice_axes->push_back(dim);
      slice_starts->push_back(start);
      slice_ends->push_back(end);
      slice_strides->push_back(step);
    }
  }
  if (!PyTuple_Check(_index)) Py_DecRef(index);
}

416
// Bind Methods
J
Jiabin Yang 已提交
417
void BindImperative(py::module *m_ptr) {
418 419
  auto &m = *m_ptr;

420 421
  BindOpFunctions(&m);

422 423
#ifndef _WIN32
  // Dygraph DataLoader signal handler
424 425 426 427 428 429 430 431 432 433 434 435 436
  m.def("_set_process_pids", [](int64_t key, py::object &obj) {
    PADDLE_ENFORCE_EQ(
        py::isinstance<py::tuple>(obj) || py::isinstance<py::list>(obj), true,
        platform::errors::InvalidArgument(
            "The subprocess ids set in DataLoader is illegal."
            "Expected data type is tuple or list, but received %s",
            obj.get_type()));
    py::list pids = py::cast<py::list>(obj);
    std::set<pid_t> pids_set = {};
    for (size_t i = 0; i < pids.size(); i++) {
      pids_set.insert(pids[i].cast<pid_t>());
    }
    imperative::SetLoadProcessPIDs(key, pids_set);
437
  });
438 439
  m.def("_erase_process_pids",
        [](int64_t key) { imperative::EraseLoadProcessPIDs(key); });
440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511
  m.def("_set_process_signal_handler",
        []() { imperative::SetLoadProcessSignalHandler(); });
  m.def("_throw_error_if_process_failed",
        []() { imperative::ThrowErrorIfLoadProcessFailed(); });

  // Dygraph DataLoader reader process & thread related functions
  m.def(
      "_convert_to_tensor_list",
      [](py::object &obj) -> py::list {
        // 0. input data check
        PADDLE_ENFORCE(
            py::isinstance<py::tuple>(obj) || py::isinstance<py::list>(obj),
            platform::errors::InvalidArgument(
                "The batch data read into DataLoader is illegal."
                "Expected data type is tuple or list, but received %s",
                obj.get_type()));
        py::list batch = py::cast<py::list>(obj);
        py::list tensors;
        for (size_t i = 0; i < batch.size(); ++i) {
          // 1. cast to python array
          auto array = batch[i].cast<py::array>();
          PADDLE_ENFORCE_NE(
              string::Sprintf("%s", array.dtype()).compare("object"), 0,
              platform::errors::InvalidArgument(
                  "Faild to convert input data to a regular ndarray.\n  * "
                  "Usually this means the input data contains nested "
                  "lists with different lengths.\n  * Check the reader "
                  "function passed to 'set_(sample/sample_list/batch)"
                  "_generator' to locate the data causes this issue."));
          // 2. construcct LoDTensor
          framework::LoDTensor t;
          SetTensorFromPyArray<platform::CPUPlace>(&t, array,
                                                   platform::CPUPlace(), true);
          // 3. allocate shared memory
          void *data_ptr = t.data<void>();
          size_t data_size = t.numel() * framework::SizeOfType(t.type());
          auto shared_writer_holder =
              memory::allocation::AllocateMemoryMapWriterAllocation(data_size);
          // 4. maintain mmap fd set & backup ipc_name
          const std::string &ipc_name = shared_writer_holder->ipc_name();
          memory::allocation::MemoryMapFdSet::Instance().Insert(ipc_name);
          // 5. copy data & reset holder
          memory::Copy(platform::CPUPlace(), shared_writer_holder->ptr(),
                       platform::CPUPlace(), data_ptr, data_size);
          t.ResetHolder(shared_writer_holder);
          // 6. append to result list
          tensors.append(t);
        }
        return tensors;
      },
      py::return_value_policy::take_ownership);

  m.def("_remove_tensor_list_mmap_fds", [](py::list &tensor_list) {
    for (size_t i = 0; i < tensor_list.size(); ++i) {
      auto t = tensor_list[i].cast<framework::LoDTensor>();
      auto *mmap_writer_allocation =
          dynamic_cast<memory::allocation::MemoryMapWriterAllocation *>(
              t.Holder().get());
      PADDLE_ENFORCE_NOT_NULL(
          mmap_writer_allocation,
          platform::errors::NotFound("The shared memory of LoDTensor in "
                                     "DataLoader's child process has been "
                                     "released."));
      memory::allocation::MemoryMapFdSet::Instance().Remove(
          mmap_writer_allocation->ipc_name());
    }
  });

  m.def("_cleanup_mmap_fds",
        []() { memory::allocation::MemoryMapFdSet::Instance().Clear(); });
#endif

512 513 514 515 516
  m.def("start_imperative_gperf_profiler",
        []() { imperative::StartProfile(); });

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

Z
Zeng Jinle 已提交
517 518 519
  m.def("_is_dygraph_debug_enabled",
        []() { return imperative::IsDebugEnabled(); });
  m.def("_dygraph_debug_level", []() { return imperative::GetDebugLevel(); });
520 521 522 523
  m.def("_switch_tracer",
        [](const std::shared_ptr<imperative::Tracer> &tracer) {
          imperative::SetCurrentTracer(tracer);
        });
Z
Zeng Jinle 已提交
524

525
  py::class_<imperative::VarBase, std::shared_ptr<imperative::VarBase>>(
526
      m, "VarBase", R"DOC()DOC")
Z
Zeng Jinle 已提交
527
      .def_static("_alive_vars", &imperative::VarBase::AliveVarNames)
J
Jiabin Yang 已提交
528
      .def("__init__",
529 530 531
           [](imperative::VarBase &self, framework::proto::VarType::Type dtype,
              const std::vector<int> &dims, const py::handle &name,
              framework::proto::VarType::Type type, bool persistable) {
532
             VLOG(4) << "Init VarBase";
533 534 535
             std::string act_name = "";
             if (!name.ptr() || name.ptr() == Py_None) {
               act_name = imperative::GetCurrentTracer()->GenerateUniqueName(
536
                   "generated_tensor");
537 538 539 540
             } else {
               act_name = name.cast<std::string>();
             }
             new (&self) imperative::VarBase(act_name);
J
Jiabin Yang 已提交
541 542 543 544 545 546 547 548 549
             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));
             }
           })
550 551
      .def("__init__", &InitVarBaseFromNumpyWithArg<platform::CPUPlace>,
           py::arg("value"), py::arg("place"), py::arg("persistable") = false,
552 553
           py::arg("zero_copy") = false, py::arg("name") = "",
           py::arg("stop_gradient") = -1)
554 555 556 557
      .def("__init__", &InitVarBaseFromNumpyWithArg<platform::XPUPlace>,
           py::arg("value"), py::arg("place"), py::arg("persistable") = false,
           py::arg("zero_copy") = false, py::arg("name") = "",
           py::arg("stop_gradient") = -1)
558 559
      .def("__init__", &InitVarBaseFromNumpyWithArg<platform::CUDAPlace>,
           py::arg("value"), py::arg("place"), py::arg("persistable") = false,
560 561
           py::arg("zero_copy") = false, py::arg("name") = "",
           py::arg("stop_gradient") = -1)
562 563
      .def("__init__", &InitVarBaseFromNumpyWithArg<platform::CUDAPinnedPlace>,
           py::arg("value"), py::arg("place"), py::arg("persistable") = false,
564 565
           py::arg("zero_copy") = false, py::arg("name") = "",
           py::arg("stop_gradient") = -1)
L
Leo Chen 已提交
566
      .def("__init__", &InitVarBaseFromNumpyWithArgDefault, py::arg("value"))
567
      .def("__init__", &InitVarBaseFromTensorWithArgDefault, py::arg("tensor"))
568
      .def("__init__", &InitVarBaseFromNumpyWithKwargs)
569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595
      .def("__setitem__",
           [](std::shared_ptr<imperative::VarBase> &self, py::handle _index,
              py::object &value_obj) {
             auto self_tensor =
                 self->MutableVar()->GetMutable<framework::LoDTensor>();
             auto self_numpy = TensorToPyArray(*self_tensor);

             if (py::isinstance<py::array>(value_obj) ||
                 py::isinstance<py::int_>(value_obj) ||
                 py::isinstance<py::float_>(value_obj)) {
               auto value_numpy = value_obj;
               self_numpy[_index] = value_numpy;
               SetTensorFromPyArray(self_tensor, self_numpy,
                                    self_tensor->place(), true);

             } else {
               auto value =
                   value_obj.cast<std::shared_ptr<imperative::VarBase>>();
               auto value_tensor =
                   value->MutableVar()->GetMutable<framework::LoDTensor>();
               auto value_numpy = TensorToPyArray(*value_tensor);

               self_numpy[_index] = value_numpy;
               SetTensorFromPyArray(self_tensor, self_numpy,
                                    self_tensor->place(), true);
             }
           })
596
      .def("__getitem__",
S
songyouwei 已提交
597
           [](std::shared_ptr<imperative::VarBase> &self, py::handle _index) {
598
             std::vector<int> slice_axes, slice_starts, slice_ends,
S
songyouwei 已提交
599 600 601 602 603 604
                 slice_strides, decrease_axis, infer_flags;
             auto tensor =
                 self->MutableVar()->GetMutable<framework::LoDTensor>();
             ParseIndexingSlice(tensor, _index.ptr(), &slice_axes,
                                &slice_starts, &slice_ends, &slice_strides,
                                &decrease_axis, &infer_flags);
605 606 607 608
             // release gil and do tracing
             py::gil_scoped_release release;
             const auto &tracer = imperative::GetCurrentTracer();
             if (slice_axes.empty()) {
S
songyouwei 已提交
609
               return self;
610
             } else {
S
songyouwei 已提交
611
               imperative::NameVarBaseMap ins = {{"Input", {self}}};
612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633
               framework::AttributeMap attrs = {
                   {"axes", slice_axes},
                   {"starts", slice_starts},
                   {"ends", slice_ends},
                   {"infer_flags", infer_flags},
                   {"decrease_axis", decrease_axis}};
               auto out = std::shared_ptr<imperative::VarBase>(
                   new imperative::VarBase(tracer->GenerateUniqueName()));
               imperative::NameVarBaseMap outs = {{"Out", {out}}};
               std::string op_type = "slice";
               for (auto stride : slice_strides) {
                 if (stride != 1) {
                   op_type = "strided_slice";
                   attrs.insert({"strides", slice_strides});
                   attrs.erase("decrease_axis");
                   break;
                 }
               }
               tracer->TraceOp(op_type, ins, outs, std::move(attrs));
               return out;
             }
           })
634 635 636 637 638 639 640
      .def("numpy",
           [](imperative::VarBase &self) -> py::array {
             const auto &tensor =
                 self.MutableVar()->Get<framework::LoDTensor>();
             PADDLE_ENFORCE_EQ(
                 tensor.IsInitialized(), true,
                 platform::errors::InvalidArgument(
641
                     "Tensor of %s is Empty, please check if it has no data.",
642 643 644 645 646
                     self.Name()));
             return TensorToPyArray(tensor, true);
           },
           R"DOC(
        **Notes**:
T
tianshuo78520a 已提交
647
            **This API is ONLY available in Dygraph mode**
648 649 650 651 652 653 654 655 656 657 658 659 660 661

        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
662
                from paddle.fluid.dygraph import Linear
663 664 665 666
                import numpy as np

                data = np.random.uniform(-1, 1, [30, 10, 32]).astype('float32')
                with fluid.dygraph.guard():
667
                    linear = Linear(32, 64)
668
                    data = to_variable(data)
669
                    x = linear(data)
670 671 672 673 674 675 676 677 678 679 680 681 682
                    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(

683
        Returns a new Tensor, detached from the current graph.
684

685
        Returns: The detached Tensor.
686 687 688 689

        Examples:
            .. code-block:: python

690 691 692 693 694
                import paddle
                linear = Linear(32, 64)
                data = paddle.uniform(shape=[30, 10, 32], -1, 1)
                x = linear(data)
                y = x.detach()
695 696 697
       )DOC")
      .def("clear_gradient", &imperative::VarBase::ClearGradient, R"DOC(

698
        Only for Tensor that has gradient, normally we use this for Parameters since other temporary Tensor doesen't has gradient.
699

700
        The Gradient of current Tensor will be set to ``0`` .
701 702 703 704 705 706

        Returns:  None

        Examples:
             .. code-block:: python

707
                import paddle
Z
Zhou Wei 已提交
708 709 710 711 712 713 714
                input = paddle.uniform([10, 2])
                linear = paddle.nn.Linear(2, 3)
                out = linear(input)
                out.backward()
                print("Before clear_gradient, linear.weight.grad: {}".format(linear.weight.grad))
                linear.weight.clear_gradient()
                print("After clear_gradient, linear.weight.grad: {}".format(linear.weight.grad))
715
      )DOC")
Z
Zhou Wei 已提交
716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763
      .def("clone",
           [](std::shared_ptr<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()));
             auto tracer = imperative::GetCurrentTracer();
             auto new_var = std::make_shared<imperative::VarBase>(
                 true, tracer->GenerateUniqueName(self->Name() + "_clone"));
             framework::AttributeMap attrs;
             imperative::NameVarBaseMap ins = {{"X", {self}}};
             imperative::NameVarBaseMap outs = {{"Out", {new_var}}};
             tracer->TraceOp("assign", ins, outs, attrs);
             return new_var;
           },
           py::return_value_policy::copy, R"DOC(

        Returns a new Tensor, which is clone of origin Tensor, and it remains in the current graph.
        It will always have a Tensor copy.
        Tn addition, the cloned Tensor provides gradient propagation.

        Returns: The cloned Tensor.

        Examples:
            .. code-block:: python

              import paddle

              x = paddle.to_tensor(1.0, stop_gradient=False)
              clone_x = x.clone()
              y = clone_x**2
              y.backward()
              print(clone_x.stop_gradient) # False
              print(clone_x.grad)          # [2.0], support gradient propagation
              print(x.stop_gradient)       # False
              print(x.grad)                # [2.0], clone_x support gradient propagation for x

              x = paddle.to_tensor(1.0)
              clone_x = x.clone()
              clone_x.stop_gradient = False
              z = clone_x**3
              z.backward()
              print(clone_x.stop_gradient) # False
              print(clone_x.grad)          # [3.0], support gradient propagation
              print(x.stop_gradient) # True
              print(x.grad)          # None
       )DOC")
L
Leo Chen 已提交
764
      .def("_run_backward",
765 766
           [](imperative::VarBase &self, const imperative::Tracer &tracer,
              bool retain_graph) {
767 768
             // TODO(jiabin): when we impl more backward execution we can
             // select them
769
             auto *engine = tracer.GetEngine();
770
             engine->Init(&self, retain_graph);
771
             VLOG(3) << "Start backward";
L
Leo Chen 已提交
772 773 774 775 776 777 778 779 780 781
             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)
782 783 784 785
      .def("_set_grad_type",
           [](imperative::VarBase &self, framework::proto::VarType::Type type) {
             self.MutableGradVarBase()->SetType(type);
           })
786
      .def("_grad_ivar",
J
Jiabin Yang 已提交
787 788
           [](const imperative::VarBase &self) {
             auto &grad_var = self.GradVarBase();
789 790 791 792 793 794 795 796 797 798 799
             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
800
             }
801
             return std::shared_ptr<imperative::VarBase>(nullptr);
J
Jiabin Yang 已提交
802 803
           },
           py::return_value_policy::copy)
804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833
      .def("_is_sparse",
           [](imperative::VarBase &self) {
             return self.Var().IsType<framework::SelectedRows>();
           })
      .def("_allreduce",
           [](imperative::VarBase &self,
              const imperative::ParallelStrategy &strategy) {
             if (strategy.nranks_ > 1) {
#ifdef PADDLE_WITH_NCCL
#if NCCL_VERSION_CODE >= 2212
               imperative::AllReduce(self.Var(), self.MutableVar(), strategy);
#else
               if (!self.Var().IsType<framework::SelectedRows>()) {
                 imperative::AllReduce(self.Var(), self.MutableVar(), strategy);
               } else {
                 PADDLE_THROW(platform::errors::Unimplemented(
                     "Imperative SelectedRows allreduce is not supported when "
                     "paddle is compiled with NCCL verison lower than v2.2.12. "
                     "You can set is_sparse=False for the Layer containing "
                     "this argument, such as Embedding(is_sparse=False)."));
               }
#endif  // NCCL_VERSION_CODE
#else
               PADDLE_THROW(platform::errors::Unimplemented(
                   "Imperative allreduce is not supported when paddle is "
                   "not compiled with NCCL."));
#endif  // PADDLE_WITH_NCCL
             }
           },
           py::call_guard<py::gil_scoped_release>())
834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954
      .def("cpu",
           [](const std::shared_ptr<imperative::VarBase> &self) {
             if (platform::is_cpu_place(self->Place())) {
               return self;
             } else {
               auto new_var = self->NewVarBase(platform::CPUPlace(), true);
               new_var->SetOverridedStopGradient(self->OverridedStopGradient());
               return new_var;
             }
           },
           R"DOC(
        Returns a copy of this Tensor in CPU memory.

        If this Tensor is already in CPU memory, then no copy is performed and the original Tensor is returned.

        Examples:
            .. code-block:: python

              import paddle
              x = paddle.to_tensor(1.0, place=paddle.CUDAPlace(0))
              print(x.place)    # CUDAPlace(0)
              
              y = x.cpu()
              print(y.place)    # CPUPlace

              )DOC")
      .def("pin_memory",
           [](const std::shared_ptr<imperative::VarBase> &self) {
#ifndef PADDLE_WITH_CUDA
             PADDLE_THROW(platform::errors::PermissionDenied(
                 "Cannot copy this Tensor to pinned memory in CPU version "
                 "Paddle, "
                 "Please recompile or reinstall Paddle with CUDA support."));
#endif
             if (platform::is_cuda_pinned_place(self->Place())) {
               return self;
             } else {
               auto new_var =
                   self->NewVarBase(platform::CUDAPinnedPlace(), true);
               new_var->SetOverridedStopGradient(self->OverridedStopGradient());
               return new_var;
             }
           },
           R"DOC(
        Returns a copy of this Tensor in pin memory.

        If this Tensor is already in pin memory, then no copy is performed and the original Tensor is returned.

        Examples:
            .. code-block:: python

              import paddle
              x = paddle.to_tensor(1.0, place=paddle.CUDAPlace(0))
              print(x.place)      # CUDAPlace(0)

              y = x.pin_memory()
              print(y.place)      # CUDAPinnedPlace

      )DOC")
      .def("cuda",
           [](const std::shared_ptr<imperative::VarBase> &self, int device_id,
              bool blocking) {
#ifndef PADDLE_WITH_CUDA
             PADDLE_THROW(platform::errors::PermissionDenied(
                 "Cannot copy this Tensor to GPU in CPU version Paddle, "
                 "Please recompile or reinstall Paddle with CUDA support."));
#else
             int device_count = platform::GetCUDADeviceCount();
             if (device_id == -1) {
               if (platform::is_gpu_place(self->Place())) {
                 return self;
               } else {
                 device_id = 0;
               }
             }
             PADDLE_ENFORCE_GE(
                 device_id, 0,
                 platform::errors::InvalidArgument(
                     "Can not copy Tensor to Invalid CUDAPlace(%d), device id "
                     "must inside [0, %d)",
                     device_id, device_count));
             PADDLE_ENFORCE_LT(
                 device_id, device_count,
                 platform::errors::InvalidArgument(
                     "Can not copy Tensor to Invalid CUDAPlace(%d), device id "
                     "must inside [0, %d)",
                     device_id, device_count));
             platform::CUDAPlace place = platform::CUDAPlace(device_id);
             if (platform::is_same_place(self->Place(), place)) {
               return self;
             } else {
               auto new_var = self->NewVarBase(place, blocking);
               new_var->SetOverridedStopGradient(self->OverridedStopGradient());
               return new_var;
             }
#endif
           },
           py::arg("device_id") = -1, py::arg("blocking") = true, R"DOC(
        Returns a copy of this Tensor in GPU memory.

        If this Tensor is already in GPU memory and device_id is default, 
        then no copy is performed and the original Tensor is returned.
        
        Args:
            device_id(int, optional): The destination GPU device id. Defaults to the current device.
            blocking(bool, optional): If False and the source is in pinned memory, the copy will be 
              asynchronous with respect to the host. Otherwise, the argument has no effect. Default: False.

        Examples:
            .. code-block:: python

              import paddle
              x = paddle.to_tensor(1.0, place=paddle.CPUPlace())
              print(x.place)        # CPUPlace

              y = x.cuda()
              print(y.place)        # CUDAPlace(0)

              y = x.cuda(1)
              print(y.place)        # CUDAPlace(1)
       )DOC")
955 956
      .def("_copy_to",
           [](const imperative::VarBase &self, const platform::CPUPlace &place,
J
Jiabin Yang 已提交
957 958
              bool blocking) { return self.NewVarBase(place, blocking); },
           py::return_value_policy::copy)
959 960 961 962 963
      .def("_copy_to",
           [](const imperative::VarBase &self,
              const platform::CUDAPinnedPlace &place,
              bool blocking) { return self.NewVarBase(place, blocking); },
           py::return_value_policy::copy)
964 965 966 967
      .def("_copy_to",
           [](const imperative::VarBase &self, const platform::XPUPlace &place,
              bool blocking) { return self.NewVarBase(place, blocking); },
           py::return_value_policy::copy)
968 969
      .def("_copy_to",
           [](const imperative::VarBase &self, const platform::CUDAPlace &place,
J
Jiabin Yang 已提交
970 971 972
              bool blocking) { return self.NewVarBase(place, blocking); },
           py::return_value_policy::copy)
      .def("value", [](imperative::VarBase &self) { return self.MutableVar(); },
973 974 975
           py::return_value_policy::reference)
      .def_property("name", &imperative::VarBase::Name,
                    &imperative::VarBase::SetName)
L
Leo Chen 已提交
976 977 978 979 980
      .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
981 982 983 984
      .def_property_readonly(
          "shape",
          [](imperative::VarBase &self) {
            if (self.Var().IsType<framework::LoDTensor>()) {
985
              return framework::vectorize<int>(
J
Jiabin Yang 已提交
986
                  self.Var().Get<framework::LoDTensor>().dims());
987 988 989
            } else if (self.Var().IsType<framework::SelectedRows>()) {
              return framework::vectorize<int>(
                  self.Var().Get<framework::SelectedRows>().value().dims());
J
Jiabin Yang 已提交
990
            } else {
991 992
              VLOG(2) << "It is meaningless to get shape of "
                         "variable type "
J
Jiabin Yang 已提交
993 994 995 996
                      << GetTypeName(self);
              return std::vector<int>();
            }
          })
997 998 999
      .def_property_readonly(
          "place", [](imperative::VarBase &self) { return self.Place(); },
          py::return_value_policy::copy)
1000 1001 1002 1003 1004 1005
      .def_property_readonly("_place_str",
                             [](imperative::VarBase &self) {
                               std::stringstream ostr;
                               ostr << self.Place();
                               return ostr.str();
                             })
J
Jiabin Yang 已提交
1006
      .def_property_readonly("type", &imperative::VarBase::Type)
L
Leo Chen 已提交
1007
      .def_property_readonly("dtype", &imperative::VarBase::DataType);
1008 1009 1010

  py::class_<imperative::Layer, Layer /* <--- trampoline*/> layer(m, "Layer");
  layer.def(py::init<>())
1011 1012 1013 1014 1015
      .def("forward",
           [](imperative::Layer &self,
              const std::vector<std::shared_ptr<imperative::VarBase>> &inputs) {
             return self.Forward(inputs);
           });
1016

1017 1018 1019 1020 1021
  py::class_<imperative::jit::ProgramDescTracer>(m, "ProgramDescTracer", "")
      .def("create_program_desc",
           &imperative::jit::ProgramDescTracer::CreateProgramDesc)
      .def("reset", &imperative::jit::ProgramDescTracer::Reset);

1022
  py::class_<imperative::Tracer, std::shared_ptr<imperative::Tracer>>(
1023
      m, "Tracer", R"DOC()DOC")
1024
      .def("__init__",
J
Jiabin Yang 已提交
1025
           [](imperative::Tracer &self) { new (&self) imperative::Tracer(); })
1026 1027 1028
      .def_property("_enable_program_desc_tracing",
                    &imperative::Tracer::IsProgramDescTracingEnabled,
                    &imperative::Tracer::SetEnableProgramDescTracing)
1029 1030
      .def_property("_enable_autocast", &imperative::Tracer::IsAutoCastEnabled,
                    &imperative::Tracer::SetEnableAutoCast)
1031 1032
      .def_property("_train_mode", &imperative::Tracer::HasGrad,
                    &imperative::Tracer::SetHasGrad)
1033 1034 1035 1036 1037 1038 1039 1040
      .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
1041
              self.SetExpectedPlace(*p);
1042 1043 1044
            } else if (py::isinstance<platform::XPUPlace>(obj)) {
              auto p = obj.cast<platform::XPUPlace *>();
              self.SetExpectedPlace(*p);
1045 1046
            } else if (py::isinstance<platform::CPUPlace>(obj)) {
              auto p = obj.cast<platform::CPUPlace *>();
L
Leo Chen 已提交
1047
              self.SetExpectedPlace(*p);
1048 1049
            } else if (py::isinstance<platform::CUDAPinnedPlace>(obj)) {
              auto p = obj.cast<platform::CUDAPinnedPlace *>();
L
Leo Chen 已提交
1050
              self.SetExpectedPlace(*p);
1051
            } else {
L
Leo Chen 已提交
1052
              PADDLE_THROW(platform::errors::InvalidArgument(
1053 1054
                  "Incompatible Place Type: supports XPUPlace, CUDAPlace, "
                  "CPUPlace, "
L
Leo Chen 已提交
1055 1056
                  "and CUDAPinnedPlace, "
                  "but got Unknown Type!"));
1057 1058
            }
          })
1059 1060 1061
      .def("_get_program_desc_tracer",
           &imperative::Tracer::GetProgramDescTracer,
           py::return_value_policy::reference)
1062
      .def("_generate_unique_name", &imperative::Tracer::GenerateUniqueName,
1063
           py::arg("key") = "dygraph_tmp")
1064 1065 1066 1067 1068
      .def(
          "_set_amp_op_list",
          [](imperative::Tracer &self,
             std::unordered_set<std::string> &allow_ops,
             std::unordered_set<std::string> &block_ops) {
1069 1070
            // NOTE(zhiqiu): The automatic conversion in pybind11 between
            // c++
1071 1072 1073 1074
            // STL and python set/list/dict involve a copy operation that
            // prevents pass-by-reference semantics, so it is ok to swap.
            // The reaseon why not directly pass
            // std::shared_ptr<std::unordered_set<std::string>>
1075 1076
            // is that pybind11 forbid shared_ptr<T> where T is not custom
            // type.
1077 1078 1079 1080 1081 1082 1083 1084 1085
            imperative::AmpOperators::Instance().GetAllowOps()->swap(allow_ops);
            imperative::AmpOperators::Instance().GetBlockOps()->swap(block_ops);
          })
      .def("_get_amp_op_list",
           [](imperative::Tracer &self) {
             return std::make_tuple(
                 *(imperative::AmpOperators::Instance().GetAllowOps()),
                 *(imperative::AmpOperators::Instance().GetBlockOps()));
           })
1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098
      .def("trace",
           [](imperative::Tracer &self, const std::string &type,
              const PyNameVarBaseMap &ins, const PyNameVarBaseMap &outs,
              framework::AttributeMap attrs, const platform::XPUPlace &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);
             }
           })
M
minqiyang 已提交
1099
      .def("trace",
J
Jiabin Yang 已提交
1100 1101 1102 1103 1104 1105
           [](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);
1106 1107
             {
               py::gil_scoped_release release;
J
Jiabin Yang 已提交
1108 1109
               self.TraceOp(type, std::move(ins_map), std::move(outs_map),
                            std::move(attrs), place, trace_backward);
1110
             }
M
minqiyang 已提交
1111
           })
J
Jiabin Yang 已提交
1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124
      .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);
             }
           });
1125 1126

  // define parallel context
1127 1128 1129
  py::class_<imperative::ParallelStrategy> parallel_strategy(
      m, "ParallelStrategy", "");
  parallel_strategy.def(py::init())
1130 1131
      .def_property(
          "nranks",
1132 1133
          [](const imperative::ParallelStrategy &self) { return self.nranks_; },
          [](imperative::ParallelStrategy &self, int nranks) {
1134 1135 1136
            self.nranks_ = nranks;
          })
      .def_property("local_rank",
1137
                    [](const imperative::ParallelStrategy &self) {
1138 1139
                      return self.local_rank_;
                    },
1140
                    [](imperative::ParallelStrategy &self, int local_rank) {
1141 1142 1143 1144
                      self.local_rank_ = local_rank;
                    })
      .def_property(
          "trainer_endpoints",
1145
          [](const imperative::ParallelStrategy &self) {
1146 1147
            return self.trainer_endpoints_;
          },
1148
          [](imperative::ParallelStrategy &self, std::vector<std::string> eps) {
1149 1150 1151
            self.trainer_endpoints_ = eps;
          })
      .def_property("current_endpoint",
1152
                    [](const imperative::ParallelStrategy &self) {
1153 1154
                      return self.current_endpoint_;
                    },
1155 1156
                    [](imperative::ParallelStrategy &self,
                       const std::string &ep) { self.current_endpoint_ = ep; });
1157 1158 1159 1160 1161 1162 1163 1164

  m.def(
      "dygraph_partial_grad",
      [](const std::vector<std::shared_ptr<imperative::VarBase>> &input_targets,
         const std::vector<std::shared_ptr<imperative::VarBase>>
             &output_targets,
         const std::vector<std::shared_ptr<imperative::VarBase>> &output_grads,
         const std::vector<std::shared_ptr<imperative::VarBase>> &no_grad_vars,
1165 1166
         const platform::Place &place, bool create_graph, bool retain_graph,
         bool allow_unused, bool only_inputs) {
Z
Zeng Jinle 已提交
1167 1168
        imperative::PartialGradEngine engine(
            input_targets, output_targets, output_grads, no_grad_vars, place,
1169
            create_graph, retain_graph, allow_unused, only_inputs);
1170 1171 1172 1173 1174
        engine.Execute();
        return engine.GetResult();
      },
      py::call_guard<py::gil_scoped_release>());

1175
#if defined(PADDLE_WITH_NCCL)
1176 1177
  py::class_<imperative::NCCLParallelContext> nccl_ctx(m,
                                                       "NCCLParallelContext");
1178 1179

  nccl_ctx
1180 1181 1182
      .def(py::init<const imperative::ParallelStrategy &,
                    const platform::CUDAPlace &>())
      .def("init", [](imperative::NCCLParallelContext &self) { self.Init(); });
1183
#endif
1184 1185 1186 1187
}

}  // namespace pybind
}  // namespace paddle