imperative.cc 71.2 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 <algorithm>
24
#include <memory>
25
#include <set>
J
Jiabin Yang 已提交
26
#include <string>
27
#include <unordered_map>
28
#include <unordered_set>
29
#include <utility>
J
Jiabin Yang 已提交
30
#include <vector>
31

32
#include "paddle/fluid/imperative/all_reduce.h"
33
#include "paddle/fluid/imperative/amp_auto_cast.h"
34
#include "paddle/fluid/imperative/basic_engine.h"
35
#include "paddle/fluid/imperative/bkcl_context.h"
36
#include "paddle/fluid/imperative/data_loader.h"
37
#include "paddle/fluid/imperative/hooks.h"
38
#include "paddle/fluid/imperative/layer.h"
J
Jiabin Yang 已提交
39
#include "paddle/fluid/imperative/nccl_context.h"
40
#include "paddle/fluid/imperative/partial_grad_engine.h"
41
#include "paddle/fluid/imperative/profiler.h"
42
#include "paddle/fluid/imperative/py_layer_fwd.h"
43
#include "paddle/fluid/imperative/reducer.h"
44
#include "paddle/fluid/imperative/tracer.h"
M
minqiyang 已提交
45
#include "paddle/fluid/imperative/type_defs.h"
46
#include "paddle/fluid/memory/allocation/mmap_allocator.h"
47
#include "paddle/fluid/pybind/op_function.h"
48
#include "paddle/fluid/pybind/pybind_boost_headers.h"
L
Leo Chen 已提交
49
#include "paddle/fluid/pybind/tensor_py.h"
50

51 52 53
namespace paddle {
namespace pybind {

54 55
namespace py = ::pybind11;

56 57 58 59
class Layer : public imperative::Layer {
 public:
  using imperative::Layer::Layer;  // Inherit constructors

60 61 62 63
  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
64
                      Forward, inputs);  // NOLINT
65 66 67
  }
};

68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126
template <typename T>
static T PyObjectCast(PyObject *obj) {
  try {
    return py::cast<T>(py::handle(obj));
  } catch (py::cast_error &) {
    PADDLE_THROW(platform::errors::InvalidArgument(
        "Python object is not type of %s", typeid(T).name()));
  }
}

class PyVariableWrapperHook : public imperative::VariableWrapperHook {
 public:
  explicit PyVariableWrapperHook(PyObject *func) : py_func_(func) {
    Py_INCREF(py_func_);
  }

  ~PyVariableWrapperHook() {
    py::gil_scoped_acquire gil;
    Py_DECREF(py_func_);
  }

  std::shared_ptr<imperative::VariableWrapper> operator()(
      const std::shared_ptr<imperative::VariableWrapper> &var) override {
    py::gil_scoped_acquire gil;
    VLOG(3) << "Call PyVariableWrapperHook for var " << var->Name();

    // 1. unpack temp VarBase from VariableWrapper
    std::shared_ptr<imperative::VarBase> tmp_varbase =
        std::make_shared<imperative::VarBase>(var);

    // 2. call hook and return
    PyObject *res = nullptr;
    try {
      res = PyObject_CallFunctionObjArgs(py_func_, py::cast(tmp_varbase).ptr(),
                                         nullptr);
    } catch (platform::EnforceNotMet &e) {
      throw std::move(e);
    } catch (std::exception &e) {
      PADDLE_THROW(platform::errors::Unavailable(
          "Hook function of Tensor raises an exception: %s.", e.what()));
    } catch (...) {
      PADDLE_THROW(platform::errors::Fatal(
          "Hook function of Tensor raises an unknown exception."));
    }

    PADDLE_ENFORCE_NOT_NULL(res,
                            platform::errors::Unavailable(
                                "Hook function of Tensor return a nullptr."));
    if (res == Py_None) {
      return var;
    }

    return PyObjectCast<std::shared_ptr<imperative::VarBase>>(res)->SharedVar();
  }

 private:
  PyObject *py_func_;
};

L
Leo Chen 已提交
127 128 129 130 131
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>();
132 133
  } else if (py::isinstance<platform::XPUPlace>(place_obj)) {
    return place_obj.cast<platform::XPUPlace>();
L
Leo Chen 已提交
134 135
  } else if (py::isinstance<platform::CUDAPinnedPlace>(place_obj)) {
    return place_obj.cast<platform::CUDAPinnedPlace>();
136 137
  } else if (py::isinstance<platform::Place>(place_obj)) {
    return place_obj.cast<platform::Place>();
L
Leo Chen 已提交
138 139
  } else {
    PADDLE_THROW(platform::errors::InvalidArgument(
140 141
        "Place should be one of "
        "Place/CPUPlace/XPUPlace/CUDAPlace/CUDAPinnedPlace"));
L
Leo Chen 已提交
142 143 144 145 146 147 148
  }
}

static void InitTensorForVarBase(imperative::VarBase *self,
                                 const py::array &array,
                                 const platform::Place place,
                                 bool persistable = false,
149 150
                                 bool zero_copy = false, std::string name = "",
                                 int stop_gradient = -1) {
L
Leo Chen 已提交
151
  if (name == "") {
152 153
    name =
        imperative::GetCurrentTracer()->GenerateUniqueName("generated_tensor");
L
Leo Chen 已提交
154
  }
155 156 157
  VLOG(5) << "Init Tensor as: / name: " << name
          << " / persistable: " << persistable << " / zero_copy: " << zero_copy
          << " / stop_gradient: " << stop_gradient;
L
Leo Chen 已提交
158
  new (self) imperative::VarBase(name);
159
  auto *tensor = self->MutableVar()->GetMutable<framework::LoDTensor>();
L
Leo Chen 已提交
160 161
  if (platform::is_cpu_place(place)) {
    SetTensorFromPyArray<platform::CPUPlace>(
162
        tensor, array, BOOST_GET_CONST(platform::CPUPlace, place), zero_copy);
163 164 165
  } 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
166 167
  } else if (platform::is_gpu_place(place)) {
    SetTensorFromPyArray<platform::CUDAPlace>(
168
        tensor, array, BOOST_GET_CONST(platform::CUDAPlace, place), zero_copy);
L
Leo Chen 已提交
169 170
  } else if (platform::is_cuda_pinned_place(place)) {
    SetTensorFromPyArray<platform::CUDAPinnedPlace>(
171 172
        tensor, array, BOOST_GET_CONST(platform::CUDAPinnedPlace, place),
        zero_copy);
173
  } else {
L
Leo Chen 已提交
174
    PADDLE_THROW(platform::errors::InvalidArgument(
175
        "Place should be one of CPUPlace/XPUPlace/CUDAPlace/CUDAPinnedPlace"));
J
Jiabin Yang 已提交
176
  }
177 178 179
  if (stop_gradient != -1) {
    self->SetOverridedStopGradient(stop_gradient);
  }
L
Leo Chen 已提交
180
  self->SetPersistable(persistable);
181 182 183 184 185 186
  self->SetType(framework::proto::VarType::LOD_TENSOR);
  self->SetDataType(tensor->type());
}

static void InitVarBaseFromNumpyWithKwargs(imperative::VarBase *self,
                                           const py::kwargs &kwargs) {
187
  VLOG(4) << "Init VarBase from kwargs: ";
188 189
  PADDLE_ENFORCE_EQ(
      kwargs.contains("value"), true,
190 191
      platform::errors::NotFound(
          "The kwargs used to create Varbase misses argument: value"));
L
Leo Chen 已提交
192 193 194 195 196 197 198 199
  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>() : "";
200 201 202
  auto stop_gradient = kwargs.contains("stop_gradient")
                           ? kwargs["stop_gradient"].cast<int>()
                           : -1;
L
Leo Chen 已提交
203 204 205
  auto default_place = imperative::GetCurrentTracer()->ExpectedPlace();
  auto place = kwargs.contains("place") ? PyObjectToPlace(kwargs["place"])
                                        : default_place;
206 207
  InitTensorForVarBase(self, array, place, persistable, zero_copy, name,
                       stop_gradient);
208
}
209

210 211 212
template <typename P>
static void InitVarBaseFromNumpyWithArg(imperative::VarBase *self,
                                        const py::array &array, const P &place,
L
Leo Chen 已提交
213 214
                                        bool persistable = false,
                                        bool zero_copy = false,
215 216 217 218 219
                                        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
220
  if (name == "") {
221 222
    name =
        imperative::GetCurrentTracer()->GenerateUniqueName("generated_tensor");
L
Leo Chen 已提交
223
  }
224 225
  VLOG(5) << "Init Tensor as: / name: " << name
          << " / persistable: " << persistable << " / zero_copy: " << zero_copy
226
          << " / stop_gradient: " << stop_gradient << " / at " << place;
L
Leo Chen 已提交
227
  new (self) imperative::VarBase(name);
228 229
  self->SetPersistable(persistable);
  auto *tensor = self->MutableVar()->GetMutable<framework::LoDTensor>();
230 231 232
  if (stop_gradient != -1) {
    self->SetOverridedStopGradient(stop_gradient);
  }
233 234 235 236 237 238
  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
239 240
                                               const py::array &array) {
  auto place = imperative::GetCurrentTracer()->ExpectedPlace();
241
  VLOG(4) << "Init VarBase from numpy at " << place;
L
Leo Chen 已提交
242
  InitTensorForVarBase(self, array, place);
243
}
244

245 246 247 248 249
static void InitVarBaseFromTensorWithArgDefault(
    imperative::VarBase *self, const framework::LoDTensor &tensor) {
  VLOG(4) << "Init VarBase";
  auto place = imperative::GetCurrentTracer()->ExpectedPlace();
  new (self) imperative::VarBase(
250
      imperative::GetCurrentTracer()->GenerateUniqueName("generated_tensor"));
251 252 253 254 255 256 257 258 259 260 261 262 263 264
  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";
  }
}

265 266 267 268 269
static std::string GetTypeName(const imperative::VarBase &var) {
  if (var.Type() == framework::proto::VarType::RAW) {
    return "RAW";
  } else if (!var.Var().IsInitialized()) {
    return "nullptr";
270
  } else {
271
    return framework::ToTypeName(var.Var().Type());
272 273
  }
}
L
Leo Chen 已提交
274

275
using PyNameVarBaseMap = std::unordered_map<std::string, py::handle>;
276 277 278 279 280 281 282 283 284 285 286 287 288

// 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;

289
  if (PyList_Check(py_obj)) {  // List of VarBase
290 291 292
    size_t len = PyList_GET_SIZE(py_obj);
    result.reserve(len);
    for (size_t i = 0; i < len; ++i) {
293 294 295
      PyObject *py_ivar = PyList_GET_ITEM(py_obj, i);
      PADDLE_ENFORCE_NOT_NULL(
          py_ivar, platform::errors::InvalidArgument("Python Object is NULL"));
296 297 298
      result.emplace_back(
          PyObjectCast<std::shared_ptr<imperative::VarBase>>(py_ivar));
    }
299
  } else if (PyTuple_Check(py_obj)) {  // Tuple of VarBase
300 301 302
    size_t len = PyTuple_GET_SIZE(py_obj);
    result.reserve(len);
    for (size_t i = 0; i < len; ++i) {
303 304 305
      PyObject *py_ivar = PyTuple_GET_ITEM(py_obj, i);
      PADDLE_ENFORCE_NOT_NULL(
          py_ivar, platform::errors::InvalidArgument("Python Object is NULL"));
306 307 308
      result.emplace_back(
          PyObjectCast<std::shared_ptr<imperative::VarBase>>(py_ivar));
    }
309 310 311
  } else {  // VarBase
    result.emplace_back(
        PyObjectCast<std::shared_ptr<imperative::VarBase>>(py_obj));
312 313 314 315 316
  }

  return result;
}

J
Jiabin Yang 已提交
317 318 319
static imperative::NameVarBaseMap ConvertToNameVarBaseMap(
    const PyNameVarBaseMap &map) {
  imperative::NameVarBaseMap result;
320 321 322 323 324 325
  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
326

327 328 329
  PADDLE_ENFORCE_EQ(
      PyErr_Occurred(), nullptr,
      platform::errors::InvalidArgument(py::str(py::handle(PyErr_Occurred()))));
330 331 332
  return result;
}

333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377
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;
378
    *start = std::max(*start, static_cast<Py_ssize_t>(0));
379 380 381 382 383 384 385 386 387 388 389 390 391
  }
  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;
392
    *stop = std::min(*stop, length);
393 394 395 396 397 398 399
  }
  if (*stop > length) return -1;
  if (*start >= length) return -1;
  if (*step == 0) return -1;
  return 0;
}

S
songyouwei 已提交
400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423
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);
424 425 426 427 428 429 430
    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
431 432
    infer_flags->push_back(1);
    int dim_len = shape[dim];
433 434
    if (PyCheckInteger(slice_item)) {
      // integer, PyLong_AsLong supports both int and long
S
songyouwei 已提交
435
      int start = static_cast<int>(PyLong_AsLong(slice_item));
H
hong 已提交
436
      auto s_t = start;
S
songyouwei 已提交
437
      start = start < 0 ? start + dim_len : start;
438
      if (start >= dim_len || start < 0) {
H
hong 已提交
439 440 441 442 443 444 445 446 447
        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
448 449 450 451 452 453
      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 {
454
      // slice item
S
songyouwei 已提交
455
      Py_ssize_t start, end, step;
456 457 458
      PySliceObject *p = reinterpret_cast<PySliceObject *>(slice_item);
      _PySlice_GetIndices(p, dim_len, &start, &end, &step);

S
songyouwei 已提交
459
      // :: or : or 0:dim_len:1
460 461 462
      if (start == 0 && end == dim_len && step == 1) {
        continue;
      }
S
songyouwei 已提交
463 464 465 466 467 468 469 470 471
      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);
}

472
// Bind Methods
J
Jiabin Yang 已提交
473
void BindImperative(py::module *m_ptr) {
474 475
  auto &m = *m_ptr;

476 477
  BindOpFunctions(&m);

478 479
#ifndef _WIN32
  // Dygraph DataLoader signal handler
480 481 482 483 484 485 486 487 488 489 490 491 492
  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);
493
  });
494 495
  m.def("_erase_process_pids",
        [](int64_t key) { imperative::EraseLoadProcessPIDs(key); });
496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547
  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);

K
Kaipeng Deng 已提交
548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580
  m.def("_array_to_share_memory_tensor",
        [](py::object &obj) {
          // 1. cast to python array
          auto array = obj.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);

          return t;
        },
        py::return_value_policy::take_ownership);

581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600
  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

601 602 603 604 605
  m.def("start_imperative_gperf_profiler",
        []() { imperative::StartProfile(); });

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

Z
Zeng Jinle 已提交
606 607 608
  m.def("_is_dygraph_debug_enabled",
        []() { return imperative::IsDebugEnabled(); });
  m.def("_dygraph_debug_level", []() { return imperative::GetDebugLevel(); });
609 610 611 612
  m.def("_switch_tracer",
        [](const std::shared_ptr<imperative::Tracer> &tracer) {
          imperative::SetCurrentTracer(tracer);
        });
Z
Zeng Jinle 已提交
613

614
  py::class_<imperative::VarBase, std::shared_ptr<imperative::VarBase>>(
615
      m, "VarBase", R"DOC()DOC")
Z
Zeng Jinle 已提交
616
      .def_static("_alive_vars", &imperative::VarBase::AliveVarNames)
617 618 619 620 621 622 623
      .def("__init__",
           [](imperative::VarBase &self) {
             std::string name =
                 imperative::GetCurrentTracer()->GenerateUniqueName(
                     "generated_tensor");
             new (&self) imperative::VarBase(name);
           })
J
Jiabin Yang 已提交
624
      .def("__init__",
625 626 627
           [](imperative::VarBase &self, framework::proto::VarType::Type dtype,
              const std::vector<int> &dims, const py::handle &name,
              framework::proto::VarType::Type type, bool persistable) {
628
             VLOG(4) << "Init VarBase";
629 630 631
             std::string act_name = "";
             if (!name.ptr() || name.ptr() == Py_None) {
               act_name = imperative::GetCurrentTracer()->GenerateUniqueName(
632
                   "generated_tensor");
633 634 635 636
             } else {
               act_name = name.cast<std::string>();
             }
             new (&self) imperative::VarBase(act_name);
J
Jiabin Yang 已提交
637 638 639 640 641 642 643 644 645
             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));
             }
           })
646 647
      .def("__init__", &InitVarBaseFromNumpyWithArg<platform::CPUPlace>,
           py::arg("value"), py::arg("place"), py::arg("persistable") = false,
648 649
           py::arg("zero_copy") = false, py::arg("name") = "",
           py::arg("stop_gradient") = -1)
650 651 652 653
      .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)
654 655
      .def("__init__", &InitVarBaseFromNumpyWithArg<platform::CUDAPlace>,
           py::arg("value"), py::arg("place"), py::arg("persistable") = false,
656 657
           py::arg("zero_copy") = false, py::arg("name") = "",
           py::arg("stop_gradient") = -1)
658 659
      .def("__init__", &InitVarBaseFromNumpyWithArg<platform::CUDAPinnedPlace>,
           py::arg("value"), py::arg("place"), py::arg("persistable") = false,
660 661
           py::arg("zero_copy") = false, py::arg("name") = "",
           py::arg("stop_gradient") = -1)
L
Leo Chen 已提交
662
      .def("__init__", &InitVarBaseFromNumpyWithArgDefault, py::arg("value"))
663
      .def("__init__", &InitVarBaseFromTensorWithArgDefault, py::arg("tensor"))
664
      .def("__init__", &InitVarBaseFromNumpyWithKwargs)
665 666 667 668 669
      .def("__setitem__",
           [](std::shared_ptr<imperative::VarBase> &self, py::handle _index,
              py::object &value_obj) {
             auto self_tensor =
                 self->MutableVar()->GetMutable<framework::LoDTensor>();
670 671 672 673
             PyObject *index_ptr = !PyTuple_Check(_index.ptr())
                                       ? PyTuple_Pack(1, _index.ptr())
                                       : _index.ptr();
             // 1. Check argumnets
674 675
             // 1.1 Check whether value obj is a tensor.
             bool value_is_tensor = true;
676
             bool parse_index = true;
677 678 679 680 681 682 683
             if (py::isinstance<py::array>(value_obj) ||
                 py::isinstance<py::int_>(value_obj) ||
                 py::isinstance<py::float_>(value_obj)) {
               value_is_tensor = false;
             }

             // 1.2 Check whether _index can be parsed.
684 685 686 687 688 689 690 691 692 693 694 695 696
             const int size = PyTuple_GET_SIZE(index_ptr);
             for (int dim = 0; dim < size; ++dim) {
               PyObject *slice_item = PyTuple_GetItem(index_ptr, dim);
               if (!(PyCheckInteger(slice_item) || PySlice_Check(slice_item))) {
                 parse_index = false;
                 break;
               }
             }

             // 2. Call op set_value to speed up if the condition is met,
             // otherwise call TensorToPyArray.
             // TODO(liym27): Try not to call TensorToPyArray because it always
             // copys data to cpu place, which reduces performance.
697
             if (parse_index && value_is_tensor) {
698
               std::vector<int> axes, starts, ends, steps, decrease_axes,
699 700
                   infer_flags;
               ParseIndexingSlice(self_tensor, index_ptr, &axes, &starts, &ends,
701 702 703 704 705 706 707 708
                                  &steps, &decrease_axes, &infer_flags);

               framework::AttributeMap attrs = {
                   {"axes", axes},
                   {"starts", starts},
                   {"ends", ends},
                   {"steps", steps},
                   {"decrease_axes", decrease_axes}};
709 710 711

               imperative::NameVarBaseMap ins = {{"Input", {self}}};
               imperative::NameVarBaseMap outs = {{"Out", {self}}};
712

713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741
               auto value_tensor =
                   value_obj.cast<std::shared_ptr<imperative::VarBase>>();
               ins.insert({"ValueTensor", {value_tensor}});

               const auto &tracer = imperative::GetCurrentTracer();
               {
                 // Release gil and do tracing
                 py::gil_scoped_release release;
                 tracer->TraceOp("set_value", ins, outs, std::move(attrs));
               }
             } else {
               auto self_numpy = TensorToPyArray(*self_tensor);

               if (value_is_tensor) {
                 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);
               } else {
                 auto value_numpy = value_obj;
                 self_numpy[_index] = value_numpy;
                 SetTensorFromPyArray(self_tensor, self_numpy,
                                      self_tensor->place(), true);
               }
742
             }
743 744 745 746
             // NOTE(liym27):
             // Increase the version of VarBase self because __setitem__ is an
             // inplace operator for the VarBase self.
             self->BumpInplaceVersion();
747
           })
748
      .def("__getitem__",
S
songyouwei 已提交
749
           [](std::shared_ptr<imperative::VarBase> &self, py::handle _index) {
750
             std::vector<int> slice_axes, slice_starts, slice_ends,
S
songyouwei 已提交
751 752 753 754 755 756
                 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);
757 758 759 760
             // release gil and do tracing
             py::gil_scoped_release release;
             const auto &tracer = imperative::GetCurrentTracer();
             if (slice_axes.empty()) {
S
songyouwei 已提交
761
               return self;
762
             } else {
S
songyouwei 已提交
763
               imperative::NameVarBaseMap ins = {{"Input", {self}}};
764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785
               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;
             }
           })
786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807
      .def("_inplace_version",
           [](imperative::VarBase &self) -> uint32_t {
             const auto &var = self.MutableVar();
             PADDLE_ENFORCE_EQ(
                 var->IsInitialized(), true,
                 platform::errors::InvalidArgument(
                     "Tensor of %s is Empty, please check if it has no data.",
                     self.Name()));
             return var->CurrentInplaceVersion();
           })
      .def("_bump_inplace_version",
           [](std::shared_ptr<imperative::VarBase> &self) {
             // NOTE(liym27): _bump_inplace_version is only used for inplace
             // operation
             self->BumpInplaceVersion();
           },
           R"DOC(
        **Notes**:
            **This API is ONLY available in Dygraph mode.**
            **This is a very low level API. Users should not use it directly. **
         Bump the version whenever the Tensor is modified through an inplace operation.
            )DOC")
808
      .def("numpy",
809

810 811 812 813 814 815
           [](imperative::VarBase &self) -> py::array {
             const auto &tensor =
                 self.MutableVar()->Get<framework::LoDTensor>();
             PADDLE_ENFORCE_EQ(
                 tensor.IsInitialized(), true,
                 platform::errors::InvalidArgument(
816
                     "Tensor of %s is Empty, please check if it has no data.",
817 818 819 820
                     self.Name()));
             return TensorToPyArray(tensor, true);
           },
           R"DOC(
Z
Zhou Wei 已提交
821 822
        Returns a numpy array shows the value of current Tensor.
        
823
        Returns:
Z
Zhou Wei 已提交
824
            ndarray: The numpy value of current Tensor.
825 826

        Returns type:
Z
Zhou Wei 已提交
827
            ndarray: dtype is same as current Tensor
828 829 830 831

        Examples:
            .. code-block:: python

Z
Zhou Wei 已提交
832
                import paddle
833 834
                import numpy as np
                data = np.random.uniform(-1, 1, [30, 10, 32]).astype('float32')
Z
Zhou Wei 已提交
835 836 837 838
                linear = paddle.nn.Linear(32, 64)
                data = paddle.to_tensor(data)
                x = linear(data)
                print(x.numpy())
839
       )DOC")
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
      .def("detach",
           [](const imperative::VarBase
                  &self) -> std::shared_ptr<imperative::VarBase> {
             PADDLE_ENFORCE_EQ(
                 self.Var().IsInitialized(), true,
                 platform::errors::InvalidArgument(
                     "Tensor %s has not been initialized!", self.Name()));

             PADDLE_ENFORCE_EQ(
                 self.Var().IsType<framework::LoDTensor>() ||
                     self.Var().IsType<framework::SelectedRows>(),
                 true,
                 platform::errors::InvalidArgument(
                     "Type of Tensor[%s] must be LoDTensor or SelectedRows!",
                     self.Name()));

             auto detach_var = std::make_shared<imperative::VarBase>(
                 true, "detach_" + self.Name());

             detach_var->SetPersistable(self.Persistable());
             detach_var->SetType(self.Type());
             detach_var->SetDataType(self.DataType());

             if (self.Var().IsType<framework::LoDTensor>()) {
               const auto &origin_tensor =
                   self.Var().Get<framework::LoDTensor>();
               PADDLE_ENFORCE_EQ(
                   origin_tensor.IsInitialized(), true,
                   platform::errors::InvalidArgument(
                       "Tensor %s has not been initialized!", self.Name()));

               auto *detach_tensor =
                   detach_var->MutableVar()->GetMutable<framework::LoDTensor>();
               detach_tensor->ShareDataWith(origin_tensor);
               // NOTE(liym27): Call ShareInplaceVersionCounterWith to share the
               // same TensorInplaceVersion, which is used to check whether
               // inplace
               // operations are correct.
               detach_tensor->ShareInplaceVersionCounterWith(origin_tensor);
             } else {
               const auto &origin_selected_rows =
                   self.Var().Get<framework::SelectedRows>();
               PADDLE_ENFORCE_EQ(
                   origin_selected_rows.value().IsInitialized(), true,
                   platform::errors::InvalidArgument(
                       "Tensor %s has not been initialized!", self.Name()));

               auto *detach_selected_rows =
                   detach_var->MutableVar()
                       ->GetMutable<framework::SelectedRows>();
               detach_selected_rows->set_height(origin_selected_rows.height());
               detach_selected_rows->set_rows(origin_selected_rows.rows());
               detach_selected_rows->mutable_value()->ShareDataWith(
                   origin_selected_rows.value());
               detach_selected_rows->mutable_value()
                   ->ShareInplaceVersionCounterWith(
                       origin_selected_rows.value());
             }
             VLOG(3) << "The detached Tensor(" << detach_var->Name()
                     << ") share data with " << self.Name();
             return detach_var;
           },
           py::return_value_policy::take_ownership, R"DOC(
903

904
        Returns a new Tensor, detached from the current graph.
Z
Zhou Wei 已提交
905 906
        It will share data with origin Tensor and always doesn't have a Tensor copy.
        In addition, the detached Tensor doesn't provide gradient propagation.
907

908
        Returns: The detached Tensor.
909 910 911 912

        Examples:
            .. code-block:: python

913
                import paddle
Z
Zhou Wei 已提交
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

                x = paddle.to_tensor(1.0, stop_gradient=False)
                detach_x = x.detach()
                detach_x[:] = 10.0
                print(x)  # Tensor(shape=[1], dtype=float32, place=CPUPlace, stop_gradient=False,
                          #        [10.])
                y = x**2
                y.backward()
                print(x.grad)         # [20.0]
                print(detach_x.grad)  # None, 'stop_gradient=True' by default

                detach_x.stop_gradient = False # Set stop_gradient to be False, supported auto-grad
                z = detach_x**3
                z.backward()

                print(x.grad)         # [20.0], detach_x is detached from x's graph, not affect each other
                print(detach_x.grad)  # [300.0], detach_x has its own graph

                # Due to sharing of data with origin Tensor, There are some unsafe operations:
                y = 2 * x
                detach_x[:] = 5.0
                y.backward() 
                # It will raise Error:
                #   one of the variables needed for gradient computation has been modified by an inplace operation.
             
939 940 941
       )DOC")
      .def("clear_gradient", &imperative::VarBase::ClearGradient, R"DOC(

942
        Only for Tensor that has gradient, normally we use this for Parameters since other temporary Tensor doesen't has gradient.
943

944
        The Gradient of current Tensor will be set to ``0`` .
945 946 947 948 949 950

        Returns:  None

        Examples:
             .. code-block:: python

951
                import paddle
Z
Zhou Wei 已提交
952 953 954 955 956 957 958
                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))
959
      )DOC")
Z
Zhou Wei 已提交
960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007
      .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
1008 1009 1010 1011 1012 1013
      .def("_grad_name", &imperative::VarBase::GradVarName)
      .def("_grad_value",
           [](imperative::VarBase &self) {
             return self.MutableGradVar()->Get<framework::LoDTensor>();
           },
           py::return_value_policy::reference)
1014 1015 1016 1017
      .def("_set_grad_type",
           [](imperative::VarBase &self, framework::proto::VarType::Type type) {
             self.MutableGradVarBase()->SetType(type);
           })
1018
      .def("_grad_ivar",
J
Jiabin Yang 已提交
1019 1020
           [](const imperative::VarBase &self) {
             auto &grad_var = self.GradVarBase();
1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031
             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
1032
             }
1033
             return std::shared_ptr<imperative::VarBase>(nullptr);
J
Jiabin Yang 已提交
1034 1035
           },
           py::return_value_policy::copy)
C
chentianyu03 已提交
1036 1037 1038 1039
      .def("_set_grad_ivar",
           [](imperative::VarBase &self, imperative::VarBase &grad) {
             self.SetGradVarBase(grad);
           })
1040 1041 1042 1043 1044 1045 1046 1047
      .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) {
1048
#if defined(PADDLE_WITH_NCCL) || defined(PADDLE_WITH_RCCL)
1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065
#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."));
1066
#endif  // PADDLE_WITH_NCCL or PADDLE_WITH_RCCL
1067 1068 1069
             }
           },
           py::call_guard<py::gil_scoped_release>())
1070 1071 1072
      .def("_register_grad_hook",
           [](imperative::VarBase &self, const py::handle &hook) {
             PADDLE_ENFORCE_EQ(
1073
                 !self.OverridedStopGradient() && self.HasGradVar(), true,
1074
                 platform::errors::InvalidArgument(
1075 1076 1077
                     "Cannot register gradient hook on a Tensor that stop "
                     "gradient or without gradient."));
             return self.GradVarBase()->AddVariableWrapperHook(
1078 1079 1080 1081 1082
                 std::make_shared<PyVariableWrapperHook>(hook.ptr()));
           })
      .def("_remove_grad_hook",
           [](imperative::VarBase &self, int64_t hook_id) {
             PADDLE_ENFORCE_EQ(
1083
                 !self.OverridedStopGradient() && self.HasGradVar(), true,
1084
                 platform::errors::InvalidArgument(
1085 1086 1087
                     "Cannot remove gradient hook on a Tensor that stop "
                     "gradient or without gradient."));
             return self.GradVarBase()->RemoveVariableWrapperHook(hook_id);
1088
           })
1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124
      .def("_register_backward_hook",
           [](imperative::VarBase &self, const py::handle &hook) {
             PADDLE_ENFORCE_EQ(
                 self.IsLeaf(), true,
                 platform::errors::InvalidArgument(
                     "Only can register backward hook for leaf Tensor."));
             PADDLE_ENFORCE_EQ(
                 !self.OverridedStopGradient() && self.HasGradVar(), true,
                 platform::errors::InvalidArgument(
                     "Cannot register backward hook on a Tensor that stop "
                     "gradient or without gradient."));
             auto py_func = PyObjectCast<std::function<void()>>(hook.ptr());
             self.GradVarBase()->AddVoidHook(
                 std::make_shared<std::function<void()>>(py_func));
           },
           R"DOC(
             Registers a backward hook for current Tensor.

             This hook will be called every time the gradient of current Tensor has been fully calculated.

             There are two differences with `_register_grad_hook`:
             1. This backward hook will be executed after the gradient accumulation completed across batchs,
                but the hook registered by `_register_grad_hook` will be executed the gradient accumulation
                completed in current batch.
             2. This backward hook function should have the following signature:

                  hook() -> None

                It requires no input and no return value.

             Args:
                 hook(function): A backward hook to be registered for Tensor.gradient

             Returns:
                 None
           )DOC")
1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152
      .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) {
1153
#if !defined(PADDLE_WITH_CUDA) && !defined(PADDLE_WITH_HIP)
1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186
             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) {
1187
#if !defined(PADDLE_WITH_CUDA) && !defined(PADDLE_WITH_HIP)
1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245
             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")
K
Kaipeng Deng 已提交
1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274
      .def("_share_memory",
           [](const std::shared_ptr<imperative::VarBase> &self) {
#ifndef _WIN32
             PADDLE_ENFORCE_EQ(
                 platform::is_cpu_place(self->Place()), true,
                 platform::errors::InvalidArgument(
                     "Sharing memory only support CPU Tensor currently"));
             // 1. get LoDTensor
             auto *t = self->MutableVar()->GetMutable<framework::LoDTensor>();
             // 2. 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);
             // 3. maintain mmap fd set & backup ipc_name
             const std::string &ipc_name = shared_writer_holder->ipc_name();
             memory::allocation::MemoryMapFdSet::Instance().Insert(ipc_name);
             // 4. copy data & reset holder
             memory::Copy(platform::CPUPlace(), shared_writer_holder->ptr(),
                          platform::CPUPlace(), data_ptr, data_size);
             t->ResetHolder(shared_writer_holder);
             return *t;
#else
             PADDLE_THROW(platform::errors::PermissionDenied(
                 "Sharing memory in Windows OS is not supported currently"));
#endif
           },
           py::return_value_policy::reference)
1275
      .def("copy_", &imperative::VarBase::CopyFrom)
1276
      .def("_copy_to",
1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292
           [](const std::shared_ptr<imperative::VarBase> &self,
              const platform::CPUPlace &place, bool blocking) {
             auto new_var = self->NewVarBase(place, blocking);
             // Note(zhiqiu): Since NewVarBase may use GpuCopyAsync to
             // copy data from the tensor of self to the tensor of new varbase,
             // we need to ensure that the varbase self is not destructed until
             // the GpuCopyAsync is completed. Otherwise, the memory may be
             // freed
             // when varbase self is destructed.
             // To do that, we increase the reference count of self by 1 and
             // add a cuda event to wait the GpuCopyAsync's completion.
             if (!blocking) {
               IncreaseVarbaseReferenceCountUntilCopyComplete(self, place);
             }
             return new_var;
           },
J
Jiabin Yang 已提交
1293
           py::return_value_policy::copy)
1294
      .def("_copy_to",
1295 1296 1297 1298 1299 1300 1301 1302
           [](const std::shared_ptr<imperative::VarBase> &self,
              const platform::CUDAPinnedPlace &place, bool blocking) {
             auto new_var = self->NewVarBase(place, blocking);
             if (!blocking) {
               IncreaseVarbaseReferenceCountUntilCopyComplete(self, place);
             }
             return new_var;
           },
1303
           py::return_value_policy::copy)
1304
      .def("_copy_to",
1305 1306 1307 1308 1309 1310 1311 1312
           [](const std::shared_ptr<imperative::VarBase> &self,
              const platform::XPUPlace &place, bool blocking) {
             auto new_var = self->NewVarBase(place, blocking);
             if (!blocking) {
               IncreaseVarbaseReferenceCountUntilCopyComplete(self, place);
             }
             return new_var;
           },
1313
           py::return_value_policy::copy)
1314
      .def("_copy_to",
1315 1316 1317 1318 1319 1320 1321 1322
           [](const std::shared_ptr<imperative::VarBase> &self,
              const platform::CUDAPlace &place, bool blocking) {
             auto new_var = self->NewVarBase(place, blocking);
             if (!blocking) {
               IncreaseVarbaseReferenceCountUntilCopyComplete(self, place);
             }
             return new_var;
           },
J
Jiabin Yang 已提交
1323
           py::return_value_policy::copy)
C
chentianyu03 已提交
1324 1325 1326 1327 1328 1329 1330 1331 1332 1333
      .def("_copy_to",
           [](const std::shared_ptr<imperative::VarBase> &self,
              const platform::Place &place, bool blocking) {
             auto new_var = self->NewVarBase(place, blocking);
             if (!blocking) {
               IncreaseVarbaseReferenceCountUntilCopyComplete(self, place);
             }
             return new_var;
           },
           py::return_value_policy::copy)
J
Jiabin Yang 已提交
1334
      .def("value", [](imperative::VarBase &self) { return self.MutableVar(); },
1335 1336 1337
           py::return_value_policy::reference)
      .def_property("name", &imperative::VarBase::Name,
                    &imperative::VarBase::SetName)
L
Leo Chen 已提交
1338 1339 1340 1341 1342
      .def_property("stop_gradient",
                    &imperative::VarBase::OverridedStopGradient,
                    &imperative::VarBase::SetOverridedStopGradient)
      .def_property("persistable", &imperative::VarBase::Persistable,
                    &imperative::VarBase::SetPersistable)
1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364
      .def_property_readonly("shape",
                             [](imperative::VarBase &self) {
                               if (self.Var().IsType<framework::LoDTensor>()) {
                                 return framework::vectorize<int>(
                                     self.Var()
                                         .Get<framework::LoDTensor>()
                                         .dims());
                               } else if (self.Var()
                                              .IsType<
                                                  framework::SelectedRows>()) {
                                 return framework::vectorize<int>(
                                     self.Var()
                                         .Get<framework::SelectedRows>()
                                         .value()
                                         .dims());
                               } else {
                                 VLOG(2) << "It is meaningless to get shape of "
                                            "variable type "
                                         << GetTypeName(self);
                                 return std::vector<int>();
                               }
                             })
1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393
      .def_property_readonly("is_leaf", &imperative::VarBase::IsLeaf,
                             R"DOC(
      Whether a Tensor is leaf Tensor.

      For the Tensor whose stop_gradient is ``True`` , it will be leaf Tensor. 
      
      For the Tensor whose stop_gradient is ``False`` , it will be leaf Tensor too if it is created by user.

      Returns:
          bool: Whether a Tensor is leaf Tensor.

      Examples:
          .. code-block:: python

              import paddle

              x = paddle.to_tensor(1.)
              print(x.is_leaf) # True

              x = paddle.to_tensor(1., stop_gradient=True)
              y = x + 1
              print(x.is_leaf) # True
              print(y.is_leaf) # True

              x = paddle.to_tensor(1., stop_gradient=False)
              y = x + 1
              print(x.is_leaf) # True
              print(y.is_leaf) # False
       )DOC")
1394 1395 1396
      .def_property_readonly(
          "place", [](imperative::VarBase &self) { return self.Place(); },
          py::return_value_policy::copy)
1397 1398 1399 1400 1401 1402
      .def_property_readonly("_place_str",
                             [](imperative::VarBase &self) {
                               std::stringstream ostr;
                               ostr << self.Place();
                               return ostr.str();
                             })
J
Jiabin Yang 已提交
1403
      .def_property_readonly("type", &imperative::VarBase::Type)
L
Leo Chen 已提交
1404
      .def_property_readonly("dtype", &imperative::VarBase::DataType);
1405 1406 1407

  py::class_<imperative::Layer, Layer /* <--- trampoline*/> layer(m, "Layer");
  layer.def(py::init<>())
1408 1409 1410 1411 1412
      .def("forward",
           [](imperative::Layer &self,
              const std::vector<std::shared_ptr<imperative::VarBase>> &inputs) {
             return self.Forward(inputs);
           });
1413

1414 1415 1416 1417 1418
  py::class_<imperative::jit::ProgramDescTracer>(m, "ProgramDescTracer", "")
      .def("create_program_desc",
           &imperative::jit::ProgramDescTracer::CreateProgramDesc)
      .def("reset", &imperative::jit::ProgramDescTracer::Reset);

1419
  py::class_<imperative::Tracer, std::shared_ptr<imperative::Tracer>>(
1420
      m, "Tracer", R"DOC()DOC")
1421
      .def("__init__",
J
Jiabin Yang 已提交
1422
           [](imperative::Tracer &self) { new (&self) imperative::Tracer(); })
1423 1424 1425
      .def_property("_enable_program_desc_tracing",
                    &imperative::Tracer::IsProgramDescTracingEnabled,
                    &imperative::Tracer::SetEnableProgramDescTracing)
1426 1427
      .def_property("_enable_autocast", &imperative::Tracer::IsAutoCastEnabled,
                    &imperative::Tracer::SetEnableAutoCast)
1428
      .def_property("_has_grad", &imperative::Tracer::HasGrad,
1429
                    &imperative::Tracer::SetHasGrad)
1430 1431 1432 1433 1434 1435 1436 1437
      .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
1438
              self.SetExpectedPlace(*p);
1439 1440
              VLOG(4) << "Tracer(" << &self << ")"
                      << " set expected place " << *p;
1441 1442 1443
            } else if (py::isinstance<platform::XPUPlace>(obj)) {
              auto p = obj.cast<platform::XPUPlace *>();
              self.SetExpectedPlace(*p);
1444 1445
              VLOG(4) << "Tracer(" << &self << ")"
                      << " set expected place " << *p;
1446 1447
            } else if (py::isinstance<platform::CPUPlace>(obj)) {
              auto p = obj.cast<platform::CPUPlace *>();
L
Leo Chen 已提交
1448
              self.SetExpectedPlace(*p);
1449 1450
              VLOG(4) << "Tracer(" << &self << ")"
                      << " set expected place " << *p;
1451 1452
            } else if (py::isinstance<platform::CUDAPinnedPlace>(obj)) {
              auto p = obj.cast<platform::CUDAPinnedPlace *>();
L
Leo Chen 已提交
1453
              self.SetExpectedPlace(*p);
1454 1455 1456 1457 1458 1459 1460
              VLOG(4) << "Tracer(" << &self << ")"
                      << " set expected place " << *p;
            } else if (py::isinstance<platform::Place>(obj)) {
              auto p = obj.cast<platform::Place *>();
              self.SetExpectedPlace(*p);
              VLOG(4) << "Tracer(" << &self << ")"
                      << " set expected place " << *p;
1461
            } else {
L
Leo Chen 已提交
1462
              PADDLE_THROW(platform::errors::InvalidArgument(
1463 1464
                  "Incompatible Place Type: supports XPUPlace, CUDAPlace, "
                  "CPUPlace, "
L
Leo Chen 已提交
1465 1466
                  "and CUDAPinnedPlace, "
                  "but got Unknown Type!"));
1467 1468
            }
          })
1469 1470 1471
      .def("_get_program_desc_tracer",
           &imperative::Tracer::GetProgramDescTracer,
           py::return_value_policy::reference)
1472
      .def("_generate_unique_name", &imperative::Tracer::GenerateUniqueName,
1473
           py::arg("key") = "dygraph_tmp")
1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492
      .def("_set_amp_op_list",
           [](imperative::Tracer &self,
              std::unordered_set<std::string> &allow_ops,
              std::unordered_set<std::string> &block_ops) {
             // NOTE(zhiqiu): The automatic conversion in pybind11 between
             // c++
             // 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>>
             // is that pybind11 forbid shared_ptr<T> where T is not custom
             // type.
             imperative::AmpOperators::Instance().GetMutableAllowOps()->swap(
                 allow_ops);
             imperative::AmpOperators::Instance().GetMutableBlockOps()->swap(
                 block_ops);
             VLOG(4) << "AMP operators changed, "
                     << imperative::AmpOperators::Instance();
           })
1493 1494 1495
      .def("_get_amp_op_list",
           [](imperative::Tracer &self) {
             return std::make_tuple(
1496 1497
                 *(imperative::AmpOperators::Instance().GetMutableAllowOps()),
                 *(imperative::AmpOperators::Instance().GetMutableBlockOps()));
1498
           })
1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511
      .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
1512
      .def("trace",
J
Jiabin Yang 已提交
1513 1514 1515 1516 1517 1518
           [](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);
1519 1520
             {
               py::gil_scoped_release release;
J
Jiabin Yang 已提交
1521 1522
               self.TraceOp(type, std::move(ins_map), std::move(outs_map),
                            std::move(attrs), place, trace_backward);
1523
             }
M
minqiyang 已提交
1524
           })
J
Jiabin Yang 已提交
1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537
      .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);
             }
           });
1538 1539

  // define parallel context
1540 1541 1542
  py::class_<imperative::ParallelStrategy> parallel_strategy(
      m, "ParallelStrategy", "");
  parallel_strategy.def(py::init())
1543 1544
      .def_property(
          "nranks",
1545 1546
          [](const imperative::ParallelStrategy &self) { return self.nranks_; },
          [](imperative::ParallelStrategy &self, int nranks) {
1547 1548 1549
            self.nranks_ = nranks;
          })
      .def_property("local_rank",
1550
                    [](const imperative::ParallelStrategy &self) {
1551 1552
                      return self.local_rank_;
                    },
1553
                    [](imperative::ParallelStrategy &self, int local_rank) {
1554 1555 1556 1557
                      self.local_rank_ = local_rank;
                    })
      .def_property(
          "trainer_endpoints",
1558
          [](const imperative::ParallelStrategy &self) {
1559 1560
            return self.trainer_endpoints_;
          },
1561
          [](imperative::ParallelStrategy &self, std::vector<std::string> eps) {
1562 1563 1564
            self.trainer_endpoints_ = eps;
          })
      .def_property("current_endpoint",
1565
                    [](const imperative::ParallelStrategy &self) {
1566 1567
                      return self.current_endpoint_;
                    },
1568
                    [](imperative::ParallelStrategy &self,
1569 1570 1571 1572 1573 1574 1575
                       const std::string &ep) { self.current_endpoint_ = ep; })
      .def_property(
          "nrings",
          [](const imperative::ParallelStrategy &self) { return self.nrings_; },
          [](imperative::ParallelStrategy &self, int nrings) {
            self.nrings_ = nrings;
          });
1576 1577 1578 1579 1580 1581 1582 1583

  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,
1584 1585
         const platform::Place &place, bool create_graph, bool retain_graph,
         bool allow_unused, bool only_inputs) {
Z
Zeng Jinle 已提交
1586 1587
        imperative::PartialGradEngine engine(
            input_targets, output_targets, output_grads, no_grad_vars, place,
1588
            create_graph, retain_graph, allow_unused, only_inputs);
1589 1590 1591 1592 1593
        engine.Execute();
        return engine.GetResult();
      },
      py::call_guard<py::gil_scoped_release>());

1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606
  m.def(
      "dygraph_run_backward",
      [](const std::vector<std::shared_ptr<imperative::VarBase>> &tensors,
         const std::vector<std::shared_ptr<imperative::VarBase>> &grad_tensors,
         bool retain_graph, const imperative::Tracer &tracer) {
        auto *engine = tracer.GetEngine();
        engine->Init(tensors, grad_tensors, retain_graph);
        VLOG(3) << "Start backward";
        engine->Execute();
        VLOG(3) << "Finish backward";
      },
      py::call_guard<py::gil_scoped_release>());

1607 1608
#if defined(PADDLE_WITH_NCCL) || defined(PADDLE_WITH_RCCL) || \
    defined(PADDLE_WITH_XPU_BKCL)
1609 1610 1611 1612 1613 1614
  py::class_<imperative::ParallelContext,
             std::shared_ptr<imperative::ParallelContext>>(m,
                                                           "ParallelContext");

  py::class_<imperative::Reducer, std::shared_ptr<imperative::Reducer>>(
      m, "Reducer", R"DOC()DOC")
S
ShenLiang 已提交
1615 1616 1617 1618 1619
      .def(py::init<const std::vector<std::shared_ptr<imperative::VarBase>> &,
                    const std::vector<std::vector<size_t>> &,
                    const std::vector<bool> &,
                    std::shared_ptr<imperative::ParallelContext>,
                    const std::vector<size_t> &, bool>())
1620
      .def("prepare_for_backward", &imperative::Reducer::PrepareForBackward,
1621
           py::arg("vars"), py::call_guard<py::gil_scoped_release>());
1622 1623 1624 1625

  m.def("assign_group_by_size", &imperative::AssignGroupBySize, py::arg("vars"),
        py::arg("is_sparse_gradient"),
        py::arg("group_size_limits") = std::vector<size_t>{25 * 1024 * 1024},
1626
        py::arg("tensor_indices") = std::vector<int64_t>{},
1627
        py::call_guard<py::gil_scoped_release>());
1628
#endif
1629

1630
#if defined(PADDLE_WITH_NCCL) || defined(PADDLE_WITH_RCCL)
1631 1632 1633 1634 1635
  py::class_<imperative::NCCLParallelContext, imperative::ParallelContext,
             std::shared_ptr<imperative::NCCLParallelContext>>(
      m, "NCCLParallelContext")
      .def(py::init<const imperative::ParallelStrategy &,
                    const platform::CUDAPlace &>())
K
kuizhiqing 已提交
1636 1637 1638 1639
      .def("init", [](imperative::NCCLParallelContext &self) { self.Init(); })
      .def("init_with_ring_id",
           &imperative::NCCLParallelContext::InitWithRingID,
           py::arg("ring_id"));
1640 1641 1642 1643 1644 1645 1646 1647
#endif

#if defined(PADDLE_WITH_XPU_BKCL)
  py::class_<imperative::BKCLParallelContext, imperative::ParallelContext,
             std::shared_ptr<imperative::BKCLParallelContext>>(
      m, "BKCLParallelContext")
      .def(py::init<const imperative::ParallelStrategy &,
                    const platform::XPUPlace &>())
K
kuizhiqing 已提交
1648 1649 1650 1651
      .def("init", [](imperative::BKCLParallelContext &self) { self.Init(); })
      .def("init_with_ring_id",
           &imperative::BKCLParallelContext::InitWithRingID,
           py::arg("ring_id"));
1652
#endif
1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675
  m.def("pylayer_apply",
        [](const platform::CPUPlace &place, const py::object &cls,
           const py::args args, const py::kwargs kwargs) {
          return imperative::PyLayerApply(place, cls, args, kwargs);
        });

  m.def("pylayer_apply",
        [](const platform::CUDAPlace &place, const py::object &cls,
           const py::args args, const py::kwargs kwargs) {
          return imperative::PyLayerApply(place, cls, args, kwargs);
        });

  m.def("pylayer_apply",
        [](const platform::XPUPlace &place, const py::object &cls,
           const py::args args, const py::kwargs kwargs) {
          return imperative::PyLayerApply(place, cls, args, kwargs);
        });

  m.def("pylayer_apply",
        [](const platform::CUDAPinnedPlace &place, const py::object &cls,
           const py::args args, const py::kwargs kwargs) {
          return imperative::PyLayerApply(place, cls, args, kwargs);
        });
1676 1677 1678 1679
}

}  // namespace pybind
}  // namespace paddle