op_desc.cc 16.8 KB
Newer Older
1
/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved.
F
fengjiayi 已提交
2 3 4 5 6 7 8 9 10 11 12 13 14

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. */

Y
Yi Wang 已提交
15
#include "paddle/fluid/framework/op_desc.h"
16
#include <algorithm>
Y
Yu Yang 已提交
17
#include <functional>
18 19
#include <mutex>  // NOLINT
#include <string>
Y
Yu Yang 已提交
20
#include <unordered_map>
21
#include "glog/logging.h"
Y
Yi Wang 已提交
22
#include "paddle/fluid/framework/block_desc.h"
Y
yuyang18 已提交
23
#include "paddle/fluid/framework/op_proto_maker.h"
Y
Yi Wang 已提交
24 25 26
#include "paddle/fluid/framework/operator.h"
#include "paddle/fluid/framework/program_desc.h"
#include "paddle/fluid/framework/shape_inference.h"
Y
Yu Yang 已提交
27

F
fengjiayi 已提交
28 29 30
namespace paddle {
namespace framework {

Y
Yu Yang 已提交
31 32
class OpDesc;
class BlockDesc;
33 34
class CompileTimeInferShapeContext : public InferShapeContext {
 public:
Y
Yu Yang 已提交
35
  CompileTimeInferShapeContext(const OpDesc &op, const BlockDesc &block);
36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52

  bool HasInput(const std::string &name) const override;

  bool HasOutput(const std::string &name) const override;

  bool HasInputs(const std::string &name) const override;

  bool HasOutputs(const std::string &name) const override;

  AttrReader Attrs() const override;

  const std::vector<std::string> &Inputs(
      const std::string &name) const override;

  const std::vector<std::string> &Outputs(
      const std::string &name) const override;

Q
Qiao Longfei 已提交
53 54 55 56 57 58
  void ShareLoD(const std::string &in, const std::string &out, size_t i = 0,
                size_t j = 0) const override {
    PADDLE_ENFORCE_LT(i, Inputs(in).size());
    PADDLE_ENFORCE_LT(j, Outputs(out).size());
    auto *in_var = block_.FindVarRecursive(Inputs(in)[i]);
    auto *out_var = block_.FindVarRecursive(Outputs(out)[j]);
59
    if (in_var->GetType() != proto::VarType::LOD_TENSOR) {
Q
Qiao Longfei 已提交
60
      VLOG(3) << "input " << in << " is not LodTensor";
Q
Qiao Longfei 已提交
61 62
      return;
    }
63
    PADDLE_ENFORCE_EQ(in_var->GetType(), proto::VarType::LOD_TENSOR,
Q
Qiao Longfei 已提交
64 65
                      "The %d-th output of Output(%s) must be LoDTensor.", j,
                      out);
66
    out_var->SetLoDLevel(in_var->GetLoDLevel());
Q
Qiao Longfei 已提交
67
  }
D
dzhwinter 已提交
68

69 70 71
  bool IsRuntime() const override;

 protected:
72
  proto::VarType::Type GetVarType(const std::string &name) const override;
Q
Qiao Longfei 已提交
73

74 75 76 77
  DDim GetDim(const std::string &name) const override;

  void SetDim(const std::string &name, const DDim &dim) override;

F
fengjiayi 已提交
78 79 80 81
  std::vector<DDim> GetRepeatedDims(const std::string &name) const override;

  void SetRepeatedDims(const std::string &name,
                       const std::vector<DDim> &dims) override;
F
fengjiayi 已提交
82

F
fengjiayi 已提交
83 84
  InferShapeVarPtr GetVarPtr(const std::string &name) override;

Y
Yu Yang 已提交
85 86
  const OpDesc &op_;
  const BlockDesc &block_;
87 88
};

Y
Yu Yang 已提交
89 90
OpDesc::OpDesc(const std::string &type, const VariableNameMap &inputs,
               const VariableNameMap &outputs, const AttributeMap &attrs) {
91
  desc_.set_type(type);
F
fengjiayi 已提交
92 93 94
  inputs_ = inputs;
  outputs_ = outputs;
  attrs_ = attrs;
F
Fix bug  
fengjiayi 已提交
95
  need_update_ = true;
F
fengjiayi 已提交
96 97
}

98
void OpDesc::CopyFrom(const OpDesc &op_desc) {
F
fengjiayi 已提交
99 100 101 102 103 104 105
  desc_.set_type(op_desc.Type());
  inputs_ = op_desc.inputs_;
  outputs_ = op_desc.outputs_;
  attrs_ = op_desc.attrs_;
  need_update_ = true;
}

106
OpDesc::OpDesc(const proto::OpDesc &desc, ProgramDesc *prog, BlockDesc *block)
107 108 109 110
    : desc_(desc), need_update_(false) {
  // restore inputs_
  int input_size = desc_.inputs_size();
  for (int i = 0; i < input_size; ++i) {
111
    const proto::OpDesc::Var &var = desc_.inputs(i);
112 113 114 115 116 117 118 119 120 121
    std::vector<std::string> &args = inputs_[var.parameter()];
    int argu_size = var.arguments_size();
    args.reserve(argu_size);
    for (int j = 0; j < argu_size; ++j) {
      args.push_back(var.arguments(j));
    }
  }
  // restore outputs_
  int output_size = desc_.outputs_size();
  for (int i = 0; i < output_size; ++i) {
122
    const proto::OpDesc::Var &var = desc_.outputs(i);
123 124 125 126 127 128 129 130
    std::vector<std::string> &args = outputs_[var.parameter()];
    int argu_size = var.arguments_size();
    args.reserve(argu_size);
    for (int j = 0; j < argu_size; ++j) {
      args.push_back(var.arguments(j));
    }
  }
  // restore attrs_
131
  for (const proto::OpDesc::Attr &attr : desc_.attrs()) {
132
    std::string attr_name = attr.name();
133
    // The sub_block referred to by the BLOCK attr hasn't been added
K
Kexin Zhao 已提交
134
    // to ProgramDesc class yet, we skip setting BLOCK attr here.
135
    if (attr.type() != proto::AttrType::BLOCK) {
136 137
      attrs_[attr_name] = GetAttrValue(attr);
    }
138
  }
139
  this->block_ = block;
140 141
}

Y
Yu Yang 已提交
142
proto::OpDesc *OpDesc::Proto() {
143
  Flush();
144
  return &desc_;
F
fengjiayi 已提交
145 146
}

Y
Yu Yang 已提交
147
const std::vector<std::string> &OpDesc::Input(const std::string &name) const {
F
fengjiayi 已提交
148 149 150 151 152 153
  auto it = inputs_.find(name);
  PADDLE_ENFORCE(it != inputs_.end(), "Input %s cannot be found in Op %s", name,
                 Type());
  return it->second;
}

Y
Yu Yang 已提交
154
std::vector<std::string> OpDesc::InputArgumentNames() const {
F
Update  
fengjiayi 已提交
155 156 157 158 159 160 161
  std::vector<std::string> retv;
  for (auto &ipt : this->inputs_) {
    retv.insert(retv.end(), ipt.second.begin(), ipt.second.end());
  }
  return retv;
}

Y
Yu Yang 已提交
162 163
void OpDesc::SetInput(const std::string &param_name,
                      const std::vector<std::string> &args) {
F
fengjiayi 已提交
164 165 166 167
  need_update_ = true;
  inputs_[param_name] = args;
}

Y
Yu Yang 已提交
168
const std::vector<std::string> &OpDesc::Output(const std::string &name) const {
F
fengjiayi 已提交
169 170 171 172 173 174
  auto it = outputs_.find(name);
  PADDLE_ENFORCE(it != outputs_.end(), "Output %s cannot be found in Op %s",
                 name, Type());
  return it->second;
}

Y
Yu Yang 已提交
175
std::vector<std::string> OpDesc::OutputArgumentNames() const {
F
Update  
fengjiayi 已提交
176 177 178 179 180 181 182
  std::vector<std::string> retv;
  for (auto &ipt : this->outputs_) {
    retv.insert(retv.end(), ipt.second.begin(), ipt.second.end());
  }
  return retv;
}

Y
Yu Yang 已提交
183 184
void OpDesc::SetOutput(const std::string &param_name,
                       const std::vector<std::string> &args) {
F
fengjiayi 已提交
185 186 187 188
  need_update_ = true;
  this->outputs_[param_name] = args;
}

Y
Yu Yang 已提交
189
proto::AttrType OpDesc::GetAttrType(const std::string &name) const {
F
fengjiayi 已提交
190 191
  auto it = attrs_.find(name);
  PADDLE_ENFORCE(it != attrs_.end(), "Attribute %s is not found", name);
192
  return static_cast<proto::AttrType>(it->second.which() - 1);
F
fengjiayi 已提交
193 194
}

Y
Yu Yang 已提交
195
std::vector<std::string> OpDesc::AttrNames() const {
F
fengjiayi 已提交
196 197 198 199 200 201 202 203
  std::vector<std::string> retv;
  retv.reserve(attrs_.size());
  for (auto &attr : attrs_) {
    retv.push_back(attr.first);
  }
  return retv;
}

Y
Yu Yang 已提交
204
void OpDesc::SetAttr(const std::string &name, const Attribute &v) {
F
fengjiayi 已提交
205 206 207 208
  this->attrs_[name] = v;
  need_update_ = true;
}

A
Abhinav Arora 已提交
209 210
void OpDesc::SetBlockAttr(const std::string &name, BlockDesc *block) {
  this->attrs_[name] = block;
F
fengjiayi 已提交
211
  need_update_ = true;
F
fengjiayi 已提交
212 213
}

Y
Yu Yang 已提交
214
void OpDesc::SetAttrMap(
F
fengjiayi 已提交
215 216 217 218 219
    const std::unordered_map<std::string, Attribute> &attr_map) {
  attrs_ = attr_map;
  need_update_ = true;
}

Y
Yu Yang 已提交
220
Attribute OpDesc::GetAttr(const std::string &name) const {
F
fengjiayi 已提交
221 222 223 224 225
  auto it = attrs_.find(name);
  PADDLE_ENFORCE(it != attrs_.end(), "Attribute %s is not found", name);
  return it->second;
}

Y
Yu Yang 已提交
226
int OpDesc::GetBlockAttr(const std::string &name) const {
F
fengjiayi 已提交
227 228
  auto it = attrs_.find(name);
  PADDLE_ENFORCE(it != attrs_.end(), "Attribute %s is not found", name);
Y
Yu Yang 已提交
229
  return boost::get<BlockDesc *>(it->second)->ID();
F
fengjiayi 已提交
230 231
}

Y
Yu Yang 已提交
232
const std::unordered_map<std::string, Attribute> &OpDesc::GetAttrMap() const {
F
fengjiayi 已提交
233 234 235
  return attrs_;
}

Y
Yu Yang 已提交
236
void OpDesc::Rename(const std::string &old_name, const std::string &new_name) {
F
Update  
fengjiayi 已提交
237
  for (auto &input : inputs_) {
F
fengjiayi 已提交
238 239
    std::replace(input.second.begin(), input.second.end(), old_name, new_name);
  }
F
Update  
fengjiayi 已提交
240 241
  for (auto &output : outputs_) {
    std::replace(output.second.begin(), output.second.end(), old_name,
F
fengjiayi 已提交
242 243 244 245 246
                 new_name);
  }
  need_update_ = true;
}

Y
Yu Yang 已提交
247 248
void OpDesc::RenameOutput(const std::string &old_name,
                          const std::string &new_name) {
Y
Yang Yang(Tony) 已提交
249 250 251 252
  for (auto &output : outputs_) {
    std::replace(output.second.begin(), output.second.end(), old_name,
                 new_name);
  }
Y
yuyang18 已提交
253 254 255 256 257 258 259

  auto it = attrs_.find(framework::OpProtoAndCheckerMaker::OpRoleVarAttrName());
  if (it != attrs_.end()) {
    auto &op_vars = boost::get<std::vector<std::string>>(it->second);
    std::replace(op_vars.begin(), op_vars.end(), old_name, new_name);
  }

Y
Yang Yang(Tony) 已提交
260 261 262
  need_update_ = true;
}

Y
Yu Yang 已提交
263 264
void OpDesc::RenameInput(const std::string &old_name,
                         const std::string &new_name) {
Y
Yang Yang(Tony) 已提交
265 266 267 268 269 270
  for (auto &input : inputs_) {
    std::replace(input.second.begin(), input.second.end(), old_name, new_name);
  }
  need_update_ = true;
}

Y
Yu Yang 已提交
271
struct SetAttrDescVisitor : public boost::static_visitor<void> {
272 273
  explicit SetAttrDescVisitor(proto::OpDesc::Attr *attr) : attr_(attr) {}
  mutable proto::OpDesc::Attr *attr_;
Y
Yu Yang 已提交
274 275 276
  void operator()(int v) const { attr_->set_i(v); }
  void operator()(float v) const { attr_->set_f(v); }
  void operator()(const std::string &v) const { attr_->set_s(v); }
Q
QI JUN 已提交
277 278 279 280 281 282 283

  // Please refer to https://github.com/PaddlePaddle/Paddle/issues/7162
  template <class T,
            class = typename std::enable_if<std::is_same<bool, T>::value>::type>
  void operator()(T b) const {
    attr_->set_b(b);
  }
Y
Yu Yang 已提交
284 285 286 287 288 289 290 291 292 293 294 295 296

  void operator()(const std::vector<int> &v) const {
    VectorToRepeated(v, attr_->mutable_ints());
  }
  void operator()(const std::vector<float> &v) const {
    VectorToRepeated(v, attr_->mutable_floats());
  }
  void operator()(const std::vector<std::string> &v) const {
    VectorToRepeated(v, attr_->mutable_strings());
  }
  void operator()(const std::vector<bool> &v) const {
    VectorToRepeated(v, attr_->mutable_bools());
  }
Q
QI JUN 已提交
297
  void operator()(BlockDesc *desc) const { attr_->set_block_idx(desc->ID()); }
298
  void operator()(int64_t v) const { attr_->set_l(v); }
Y
Yu Yang 已提交
299 300 301
  void operator()(boost::blank) const { PADDLE_THROW("Unexpected branch"); }
};

Y
Yu Yang 已提交
302
void OpDesc::Flush() {
F
fengjiayi 已提交
303
  if (need_update_) {
304
    this->desc_.mutable_inputs()->Clear();
F
fengjiayi 已提交
305
    for (auto &ipt : inputs_) {
306
      auto *input = desc_.add_inputs();
F
fengjiayi 已提交
307 308 309 310
      input->set_parameter(ipt.first);
      VectorToRepeated(ipt.second, input->mutable_arguments());
    }

311
    this->desc_.mutable_outputs()->Clear();
F
fengjiayi 已提交
312
    for (auto &opt : outputs_) {
313
      auto *output = desc_.add_outputs();
F
fengjiayi 已提交
314 315 316 317
      output->set_parameter(opt.first);
      VectorToRepeated(opt.second, output->mutable_arguments());
    }

318
    this->desc_.mutable_attrs()->Clear();
F
fengjiayi 已提交
319
    for (auto &attr : attrs_) {
320
      auto *attr_desc = desc_.add_attrs();
F
fengjiayi 已提交
321 322
      attr_desc->set_name(attr.first);
      attr_desc->set_type(
323
          static_cast<proto::AttrType>(attr.second.which() - 1));
Y
Yu Yang 已提交
324 325
      SetAttrDescVisitor visitor(attr_desc);
      boost::apply_visitor(visitor, attr.second);
F
fengjiayi 已提交
326 327 328 329 330
    }

    need_update_ = false;
  }
}
Y
Yu Yang 已提交
331

332 333 334 335 336 337 338 339 340 341
static std::once_flag init_infer_shape_funcs;

static void InitInferShapeFuncs() {
  std::call_once(init_infer_shape_funcs, [] {
    auto &map = OpInfoMap::Instance();
    auto &info_map = *map.mutable_map();

    for (auto &kern_pair : OperatorWithKernel::AllOpKernels()) {
      auto op_type = kern_pair.first;
      auto &op_info = info_map.at(op_type);
Y
Yiqun Liu 已提交
342 343
      auto op = static_cast<OperatorWithKernel *>(op_info.Creator()(
          "", VariableNameMap{}, VariableNameMap{}, AttributeMap{}));
344 345 346 347 348 349
      if (op_info.infer_shape_) {  // infer_shape has been registered.
        continue;
      }
      op_info.infer_shape_ = [op](InferShapeContext *ctx) {
        op->InferShape(ctx);
      };
Y
Yu Yang 已提交
350
    }
351
  });
Y
Yu Yang 已提交
352 353
}

Y
Yu Yang 已提交
354
void OpDesc::CheckAttrs() {
F
fengjiayi 已提交
355 356
  PADDLE_ENFORCE(!Type().empty(),
                 "CheckAttr() can not be called before type is setted.");
Y
Yu Yang 已提交
357 358 359 360 361 362
  auto *checker = OpInfoMap::Instance().Get(Type()).Checker();
  if (checker == nullptr) {
    // checker is not configured. That operator could be generated by Paddle,
    // not by users.
    return;
  }
F
fengjiayi 已提交
363 364 365
  checker->Check(attrs_);
}

Y
Yu Yang 已提交
366
void OpDesc::InferShape(const BlockDesc &block) const {
Y
Yu Yang 已提交
367
  VLOG(3) << "CompileTime infer shape on " << Type();
368 369 370 371
  InitInferShapeFuncs();
  auto &infer_shape = OpInfoMap::Instance().Get(this->Type()).infer_shape_;
  PADDLE_ENFORCE(static_cast<bool>(infer_shape),
                 "%s's infer_shape has not been registered", this->Type());
Y
Yu Yang 已提交
372
  CompileTimeInferShapeContext ctx(*this, block);
Y
Yu Yang 已提交
373 374 375 376 377 378 379 380 381 382 383 384 385
  if (VLOG_IS_ON(10)) {
    std::ostringstream sout;
    auto inames = this->InputArgumentNames();
    sout << " From [";
    std::copy(inames.begin(), inames.end(),
              std::ostream_iterator<std::string>(sout, ", "));
    sout << "] to [";
    auto onames = this->OutputArgumentNames();
    std::copy(onames.begin(), onames.end(),
              std::ostream_iterator<std::string>(sout, ", "));
    sout << "]";
    VLOG(10) << sout.str();
  }
386
  infer_shape(&ctx);
Y
Yu Yang 已提交
387 388
}

Y
Yu Yang 已提交
389
void OpDesc::InferVarType(BlockDesc *block) const {
Y
Yu Yang 已提交
390 391 392 393 394
  auto &info = OpInfoMap::Instance().Get(this->Type());
  if (info.infer_var_type_) {
    info.infer_var_type_(*this, block);
  } else {
    // all output type is LoDTensor by default
Y
Yu Yang 已提交
395 396 397
    VLOG(10) << this->Type()
             << " has not registered InferVarType. Set output variables to "
                "LOD_TENSOR";
Y
Yu Yang 已提交
398 399
    for (auto &out_pair : this->outputs_) {
      for (auto &out_var_name : out_pair.second) {
Y
Yang Yang(Tony) 已提交
400
        block->FindRecursiveOrCreateVar(out_var_name)
401
            .SetType(proto::VarType::LOD_TENSOR);
Y
Yu Yang 已提交
402 403 404 405 406
      }
    }
  }
}

407
CompileTimeInferShapeContext::CompileTimeInferShapeContext(
Y
Yu Yang 已提交
408
    const OpDesc &op, const BlockDesc &block)
409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475
    : op_(op), block_(block) {}

bool CompileTimeInferShapeContext::HasInput(const std::string &name) const {
  const std::vector<std::string> &input_names = op_.Input(name);
  auto length = input_names.size();
  if (length == 0) {
    return false;
  }
  PADDLE_ENFORCE_EQ(length, 1UL,
                    "Input(%s) should have only one value, "
                    "but it have %d now",
                    name, length);
  return block_.HasVarRecursive(input_names[0]);
}

bool CompileTimeInferShapeContext::HasOutput(const std::string &name) const {
  const std::vector<std::string> &output_names = op_.Output(name);
  auto length = output_names.size();
  if (length == 0) {
    return false;
  }
  PADDLE_ENFORCE_EQ(length, 1UL,
                    "Output(%s) should have only one value, "
                    "but it have %d now",
                    name, length);
  return block_.HasVarRecursive(output_names[0]);
}

bool CompileTimeInferShapeContext::HasInputs(const std::string &name) const {
  const std::vector<std::string> &input_names = op_.Input(name);
  if (input_names.empty()) {
    return false;
  }
  for (auto &input : input_names) {
    if (!block_.HasVarRecursive(input)) return false;
  }
  return true;
}

bool CompileTimeInferShapeContext::HasOutputs(const std::string &name) const {
  const std::vector<std::string> &output_names = op_.Output(name);
  if (output_names.empty()) {
    return false;
  }
  for (auto &output : output_names) {
    if (!block_.HasVarRecursive(output)) return false;
  }
  return true;
}

AttrReader CompileTimeInferShapeContext::Attrs() const {
  return AttrReader(op_.GetAttrMap());
}

const std::vector<std::string> &CompileTimeInferShapeContext::Inputs(
    const std::string &name) const {
  return op_.Input(name);
}

const std::vector<std::string> &CompileTimeInferShapeContext::Outputs(
    const std::string &name) const {
  return op_.Output(name);
}

DDim CompileTimeInferShapeContext::GetDim(const std::string &name) const {
  auto var = block_.FindVarRecursive(name);
  PADDLE_ENFORCE(var != nullptr, "Cannot find variable %s", name);
F
fengjiayi 已提交
476
  DDim res;
Y
Yang Yang(Tony) 已提交
477
  try {
F
fengjiayi 已提交
478
    auto shape = var->GetShape();
F
fengjiayi 已提交
479
    res = shape.empty() ? make_ddim({0UL}) : make_ddim(shape);
Y
Yang Yang(Tony) 已提交
480 481 482 483
  } catch (...) {
    VLOG(5) << "GetDim of variable " << name << " error";
    std::rethrow_exception(std::current_exception());
  }
F
fengjiayi 已提交
484 485 486
  return res;
}

F
fengjiayi 已提交
487
std::vector<DDim> CompileTimeInferShapeContext::GetRepeatedDims(
F
fengjiayi 已提交
488 489 490 491 492 493 494 495 496 497 498 499 500 501
    const std::string &name) const {
  auto var = block_.FindVarRecursive(name);
  PADDLE_ENFORCE(var != nullptr, "Cannot find variable %s", name);
  std::vector<DDim> res;
  try {
    auto shapes = var->GetShapes();
    for (const auto &s : shapes) {
      res.push_back(s.empty() ? make_ddim({0UL}) : make_ddim(s));
    }
  } catch (...) {
    VLOG(5) << "GetRepeatedDim of variable " << name << " error.";
    std::rethrow_exception(std::current_exception());
  }
  return res;
502 503 504 505
}

void CompileTimeInferShapeContext::SetDim(const std::string &name,
                                          const DDim &dim) {
F
fengjiayi 已提交
506
  block_.FindVarRecursive(name)->SetShape(vectorize(dim));
507
}
F
fengjiayi 已提交
508 509 510 511 512 513 514 515

void CompileTimeInferShapeContext::SetRepeatedDims(
    const std::string &name, const std::vector<DDim> &dims) {
  auto var = block_.FindVarRecursive(name);
  PADDLE_ENFORCE(var != nullptr, "Cannot find variable %s", name);
  std::vector<std::vector<int64_t>> dim_vec(dims.size());
  std::transform(dims.begin(), dims.end(), dim_vec.begin(), vectorize);
  var->SetShapes(dim_vec);
516
}
F
fengjiayi 已提交
517

518 519
bool CompileTimeInferShapeContext::IsRuntime() const { return false; }

520
proto::VarType::Type CompileTimeInferShapeContext::GetVarType(
521 522 523
    const std::string &name) const {
  return block_.FindVarRecursive(name)->GetType();
}
524

F
fengjiayi 已提交
525 526 527 528 529
InferShapeVarPtr CompileTimeInferShapeContext::GetVarPtr(
    const std::string &name) {
  return block_.FindVarRecursive(name);
}

F
fengjiayi 已提交
530 531
}  // namespace framework
}  // namespace paddle