graph_pattern_detector.cc 55.4 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14
// 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.

15
#include <algorithm>
Q
Qiao Longfei 已提交
16
#include <array>
17 18 19 20
#include <string>
#include <vector>

#include "paddle/fluid/framework/ir/graph_helper.h"
21
#include "paddle/fluid/framework/ir/graph_pattern_detector.h"
22
#include "paddle/fluid/framework/ir/graph_traits.h"
23
#include "paddle/fluid/framework/ir/graph_viz_pass.h"
C
chengduo 已提交
24
#include "paddle/fluid/framework/operator.h"
25
#include "paddle/fluid/platform/enforce.h"
Y
Yan Chunwei 已提交
26
#include "paddle/fluid/string/pretty_log.h"
Y
Yan Chunwei 已提交
27
#include "paddle/fluid/string/printf.h"
28

29 30 31 32
namespace paddle {
namespace framework {
namespace ir {

Y
Yan Chunwei 已提交
33 34 35 36
using string::PrettyLogEndl;
using string::PrettyLog;
using string::Style;

37 38
size_t PDPattern::id_ = 0UL;

C
chengduo 已提交
39
PDNode *PDPattern::NewNode(const std::string &name) {
Y
Yan Chunwei 已提交
40
  if (!name.empty()) {
T
tensor-tang 已提交
41
    PADDLE_ENFORCE_EQ(node_map_.count(name), 0UL,
Y
Yan Chunwei 已提交
42 43 44 45 46
                      "PDNode's name should be unique, get duplicate [%s]",
                      name);
  }

  nodes_.emplace_back(new PDNode(this, name));
C
chengduo 已提交
47
  auto *cur = nodes_.back().get();
Y
Yan Chunwei 已提交
48 49 50 51
  node_map_[name] = cur;
  return cur;
}

C
chengduo 已提交
52
PDNode *PDPattern::NewNode(PDNode::teller_t &&teller, const std::string &name) {
53
  if (!name.empty()) {
T
tensor-tang 已提交
54
    PADDLE_ENFORCE_EQ(node_map_.count(name), 0UL,
55 56 57 58
                      "PDNode's name should be unique, get duplicate [%s]",
                      name);
  }

59
  nodes_.emplace_back(new PDNode(std::move(teller), this, name));
C
chengduo 已提交
60
  auto *cur = nodes_.back().get();
61
  node_map_[name] = cur;
62 63 64
  return cur;
}

C
chengduo 已提交
65
PDNode *PDPattern::RetrieveNode(const std::string &id) const {
66 67 68 69 70 71 72 73
  auto it = node_map_.find(id);
  if (it == node_map_.end()) {
    return nullptr;
  }

  return it->second;
}

C
chengduo 已提交
74
void PDPattern::AddEdge(PDNode *a, PDNode *b) {
75 76 77 78 79 80
  PADDLE_ENFORCE(a);
  PADDLE_ENFORCE(b);
  PADDLE_ENFORCE(a != b, "can't connect to the same nodes.");
  edges_.emplace_back(a, b);
}

C
chengduo 已提交
81
void GraphPatternDetector::operator()(Graph *graph,
82
                                      GraphPatternDetector::handle_t handler) {
83 84 85 86
  if (!MarkPDNodesInGraph(*graph)) {
    return;
  }

87 88 89
  auto subgraphs = DetectPatterns();
  UniquePatterns(&subgraphs);
  RemoveOverlappedMatch(&subgraphs);
Y
Yan Chunwei 已提交
90
  ValidateByNodeRole(&subgraphs);
91

Y
Yan Chunwei 已提交
92
  if (subgraphs.empty()) return;
93 94
  PrettyLogEndl(Style::detail(), "---  detected %d subgraphs",
                subgraphs.size());
95
  int id = 0;
C
chengduo 已提交
96
  for (auto &g : subgraphs) {
M
minqiyang 已提交
97
    VLOG(3) << "optimizing #" << id++ << " subgraph";
98 99 100 101
    handler(g, graph);
  }
}

C
chengduo 已提交
102
bool GraphPatternDetector::MarkPDNodesInGraph(const ir::Graph &graph) {
M
minqiyang 已提交
103
  VLOG(3) << "mark pdnodes in graph";
104 105
  if (graph.Nodes().empty()) return false;

C
chengduo 已提交
106 107
  for (auto &node : GraphTraits::DFS(graph)) {
    for (const auto &pdnode : pattern_.nodes()) {
108
      if (pdnode->Tell(&node)) {
109
        VLOG(4) << "Node " << node.Name() << " marked as " << pdnode->name();
110 111 112 113
        pdnodes2nodes_[pdnode.get()].insert(&node);
      }
    }
  }
Y
Yan Chunwei 已提交
114
  // Check to early stop if some PDNode can't find matched Node.
C
chengduo 已提交
115
  for (auto &pdnode : pattern_.nodes()) {
Y
Yan Chunwei 已提交
116
    if (!pdnodes2nodes_.count(pdnode.get())) {
M
minqiyang 已提交
117
      VLOG(4) << pdnode->name() << " can't find matched Node, early stop";
Y
Yan Chunwei 已提交
118
      // return false;
Y
Yan Chunwei 已提交
119 120
    }
  }
M
minqiyang 已提交
121
  VLOG(3) << pdnodes2nodes_.size() << " nodes marked";
122

123 124 125
  return !pdnodes2nodes_.empty();
}

Y
Yan Chunwei 已提交
126 127 128
// The intermediate Nodes can only link to the nodes inside the pattern, or this
// subgraph will be droped.
void GraphPatternDetector::ValidateByNodeRole(
C
chengduo 已提交
129
    std::vector<GraphPatternDetector::subgraph_t> *subgraphs) {
Y
Yan Chunwei 已提交
130 131 132 133 134
  std::vector<GraphPatternDetector::subgraph_t> result;

  subgraphs->erase(
      std::remove_if(
          subgraphs->begin(), subgraphs->end(),
C
chengduo 已提交
135
          [](const GraphPatternDetector::subgraph_t &subgraph) -> bool {
Y
Yan Chunwei 已提交
136
            // Collect the inputs and outputs.
C
chengduo 已提交
137 138
            std::unordered_set<Node *> ios;
            for (auto &item : subgraph) {
Y
Yan Chunwei 已提交
139 140 141 142
              if (!item.first->IsIntermediate()) {
                ios.insert(item.second);
              }
            }
C
chengduo 已提交
143
            for (auto &item : subgraph) {
Y
Yan Chunwei 已提交
144
              if (item.first->IsIntermediate()) {
C
chengduo 已提交
145
                for (auto *x : item.second->inputs) {
Y
Yan Chunwei 已提交
146 147 148 149
                  if (!ios.count(x)) {
                    return true;
                  }
                }
C
chengduo 已提交
150
                for (auto *x : item.second->outputs) {
Y
Yan Chunwei 已提交
151 152 153 154 155 156 157 158 159 160 161
                  if (!ios.count(x)) {
                    return true;
                  }
                }
              }
            }
            return false;
          }),
      subgraphs->end());
}

162
struct HitGroup {
C
chengduo 已提交
163
  std::unordered_map<PDNode *, Node *> roles;
164

C
chengduo 已提交
165
  bool Match(Node *node, PDNode *pat) {
166
    if (nodes_.count(node)) {
T
Tao Luo 已提交
167 168 169 170 171
      if (roles.count(pat) && roles[pat] == node) return true;
      return false;
    } else {
      if (roles.count(pat) && roles[pat] != node) return false;
      return true;
172
    }
173 174
  }

C
chengduo 已提交
175
  void Register(Node *node, PDNode *pat) {
176 177 178 179 180
    roles[pat] = node;
    nodes_.insert(node);
  }

 private:
C
chengduo 已提交
181
  std::unordered_set<Node *> nodes_;
182 183 184
};

// Tell whether Node a links to b.
C
chengduo 已提交
185 186
bool IsNodesLink(Node *a, Node *b) {
  for (auto *node : a->outputs) {
187 188 189 190 191 192 193
    if (b == node) {
      return true;
    }
  }
  return false;
}

194 195
std::vector<GraphPatternDetector::subgraph_t>
GraphPatternDetector::DetectPatterns() {
196
  // Init empty subgraphs.
197
  std::vector<GraphPatternDetector::subgraph_t> result;
198
  std::vector<HitGroup> init_groups;
199
  std::array<std::vector<HitGroup>, 2> bi_records;
C
chengduo 已提交
200
  auto *first_pnode = pattern_.edges().empty() ? pattern().nodes().front().get()
201
                                               : pattern_.edges().front().first;
202
  if (!pdnodes2nodes_.count(first_pnode)) return result;
C
chengduo 已提交
203
  for (auto *node : pdnodes2nodes_[first_pnode]) {
204 205 206 207 208 209 210 211 212 213
    HitGroup group;
    group.roles[first_pnode] = node;
    init_groups.emplace_back(group);
  }

  int step = 0;
  bi_records[0] = std::move(init_groups);

  // Extend a PDNode to subgraphs by deducing the connection relations defined
  // in edges of PDNodes.
C
chengduo 已提交
214
  for (const auto &edge : pattern_.edges()) {
M
minqiyang 已提交
215
    VLOG(4) << "check " << edge.first->name() << " -> " << edge.second->name();
Y
Yan Chunwei 已提交
216
    // TODO(Superjomn) Fix bug here, the groups might be duplicate here.
217 218
    // Each role has two PDNodes, which indicates two roles.
    // Detect two Nodes that can match these two roles and they are connected.
C
chengduo 已提交
219 220
    auto &pre_groups = bi_records[step % 2];
    auto &cur_groups = bi_records[1 - (step++ % 2)];
221
    cur_groups.clear();
222
    if (pre_groups.empty()) break;
223
    // source -> target
C
chengduo 已提交
224 225
    for (Node *source : pdnodes2nodes_[edge.first]) {
      for (Node *target : pdnodes2nodes_[edge.second]) {
M
minqiyang 已提交
226
        VLOG(8) << "check " << source->id() << " -- " << target->id();
227
        // TODO(Superjomn) add some prune strategies.
C
chengduo 已提交
228
        for (const auto &group : pre_groups) {
T
Tao Luo 已提交
229 230 231 232 233 234
          if (IsNodesLink(source, target)) {
            HitGroup new_group = group;
            bool flag = new_group.Match(source, edge.first) &&
                        new_group.Match(target, edge.second);
            if (flag) {
              new_group.Register(source, edge.first);
235 236 237 238 239 240 241 242
              new_group.Register(target, edge.second);
              cur_groups.push_back(new_group);
              // TODO(Superjomn) need to unique
            }
          }
        }
      }
    }
M
minqiyang 已提交
243
    VLOG(3) << "step " << step << " get records: " << cur_groups.size();
C
chengduo 已提交
244 245
    for (auto &group : cur_groups) {
      for (auto &item : group.roles) {
M
minqiyang 已提交
246
        VLOG(4) << "node " << item.second->id() << " as " << item.first->name();
Y
Yan Chunwei 已提交
247
      }
M
minqiyang 已提交
248
      VLOG(4) << "=========================================================";
Y
Yan Chunwei 已提交
249
    }
250 251
  }

C
chengduo 已提交
252
  for (auto &group : bi_records[step % 2]) {
253
    GraphPatternDetector::subgraph_t subgraph;
C
chengduo 已提交
254
    for (auto &role : group.roles) {
255 256 257 258 259 260 261
      subgraph.emplace(role.first, role.second);
    }
    result.emplace_back(subgraph);
  }
  return result;
}

Y
Yan Chunwei 已提交
262 263
struct GraphItemLessThan {
  bool operator()(const std::pair<PDNode *, Node *> &a,
Y
Yan Chunwei 已提交
264
                  const std::pair<PDNode *, Node *> &b) {
Y
Yan Chunwei 已提交
265 266 267 268 269
    if (a.first != b.first) {
      return a.first < b.first;
    } else {
      return a.second < b.second;
    }
Y
Yan Chunwei 已提交
270
  }
Y
Yan Chunwei 已提交
271
};
Y
Yan Chunwei 已提交
272

273 274
// TODO(Superjomn) enhance the function as it marks unique unique as duplicates
// see https://github.com/PaddlePaddle/Paddle/issues/13550
275
void GraphPatternDetector::UniquePatterns(
C
chengduo 已提交
276
    std::vector<GraphPatternDetector::subgraph_t> *subgraphs) {
277
  if (subgraphs->empty()) return;
278
  std::vector<GraphPatternDetector::subgraph_t> result;
279 280

  std::unordered_set<size_t> set;
Y
Yan Chunwei 已提交
281
  std::hash<std::string> hasher;
C
chengduo 已提交
282
  for (auto &g : *subgraphs) {
Y
Yan Chunwei 已提交
283 284
    // Sort the items in the sub-graph, and transform to a string key.
    std::vector<std::pair<PDNode *, Node *>> sorted_keys(g.begin(), g.end());
Y
Yan Chunwei 已提交
285
    std::sort(sorted_keys.begin(), sorted_keys.end(), GraphItemLessThan());
Y
Yan Chunwei 已提交
286 287 288
    std::stringstream ss;
    for (auto &item : sorted_keys) {
      ss << item.first << ":" << item.second;
289
    }
Y
Yan Chunwei 已提交
290
    auto key = hasher(ss.str());
291 292 293 294 295 296 297 298
    if (!set.count(key)) {
      result.emplace_back(g);
      set.insert(key);
    }
  }
  *subgraphs = result;
}

299
void GraphPatternDetector::RemoveOverlappedMatch(
C
chengduo 已提交
300
    std::vector<subgraph_t> *subgraphs) {
301
  std::vector<subgraph_t> result;
C
chengduo 已提交
302
  std::unordered_set<Node *> node_set;
303

C
chengduo 已提交
304
  for (const auto &subgraph : *subgraphs) {
305
    bool valid = true;
C
chengduo 已提交
306
    for (auto &item : subgraph) {
Y
Yan Chunwei 已提交
307
      if (item.first->IsIntermediate() && node_set.count(item.second)) {
308 309 310 311 312
        valid = false;
        break;
      }
    }
    if (valid) {
C
chengduo 已提交
313
      for (auto &item : subgraph) {
314 315 316 317 318 319 320 321
        node_set.insert(item.second);
      }
      result.push_back(subgraph);
    }
  }
  *subgraphs = result;
}

322 323 324 325 326
std::string PDPattern::DotString() const {
  using inference::analysis::Dot;
  Dot dot;
  int id = 0;
  // Create Nodes
C
chengduo 已提交
327 328
  std::unordered_map<PDNode *, std::string> node2dot;
  for (const auto &node : nodes()) {
329 330 331 332 333
    std::string node_id = "Node" + std::to_string(id++);
    dot.AddNode(node_id, {}, node->name());
    node2dot[node.get()] = node_id;
  }
  // Create Edges
C
chengduo 已提交
334
  for (const auto &edge : edges()) {
335 336 337 338
    if (!node2dot.count(edge.first) || !node2dot.count(edge.second)) {
      LOG(ERROR) << "no node " << edge.first << " " << edge.second;
      continue;
    }
C
chengduo 已提交
339 340
    auto &src = node2dot.at(edge.first);
    auto &trg = node2dot.at(edge.second);
341 342 343 344 345
    dot.AddEdge(src, trg, {});
  }
  return dot.Build();
}

C
chengduo 已提交
346
PDNode &PDNode::LinksTo(const std::vector<PDNode *> &others) {
347
  // extend outlinks.
C
chengduo 已提交
348
  for (PDNode *x : others) {
349 350 351 352 353
    pattern_->AddEdge(this, x);
  }
  return *this;
}

C
chengduo 已提交
354
PDNode &PDNode::LinksFrom(const std::vector<PDNode *> &others) {
355
  // extend outlinks.
C
chengduo 已提交
356
  for (PDNode *x : others) {
357 358 359 360 361
    pattern_->AddEdge(x, this);
  }
  return *this;
}

C
chengduo 已提交
362 363
PDNode *PDNode::assert_is_op() {
  asserts_.emplace_back([](Node *x) { return x && x->IsOp(); });
Y
Yan Chunwei 已提交
364 365
  return this;
}
C
chengduo 已提交
366 367 368

PDNode *PDNode::assert_is_op(const std::string &op_type) {
  asserts_.emplace_back([op_type](Node *x) {
Y
Yan Chunwei 已提交
369 370 371 372
    return x && x->IsOp() && x->Op()->Type() == op_type;
  });
  return this;
}
C
chengduo 已提交
373 374 375 376 377 378 379 380

PDNode *PDNode::assert_is_var() {
  asserts_.emplace_back([](Node *x) { return x && x->IsVar(); });
  return this;
}

PDNode *PDNode::assert_is_not_ctrl_var() {
  asserts_.emplace_back([](Node *x) { return x && !x->IsCtrlVar(); });
Y
Yan Chunwei 已提交
381 382
  return this;
}
C
chengduo 已提交
383 384

PDNode *PDNode::assert_var_not_persistable() {
Y
Yan Chunwei 已提交
385
  assert_is_var();
C
chengduo 已提交
386
  asserts_.emplace_back([](Node *x) { return !x->Var()->Persistable(); });
Y
Yan Chunwei 已提交
387 388
  return this;
}
C
chengduo 已提交
389 390

PDNode *PDNode::assert_is_persistable_var() {
Y
Yan Chunwei 已提交
391
  assert_is_var();
C
chengduo 已提交
392
  asserts_.emplace_back([=](Node *x) { return x->Var()->Persistable(); });
Y
Yan Chunwei 已提交
393 394
  return this;
}
C
chengduo 已提交
395 396 397

PDNode *PDNode::assert_is_op_nth_input(const std::string &op_type,
                                       const std::string &argument, int nth) {
Y
Yan Chunwei 已提交
398 399
  assert_is_var();
  assert_is_op_input(op_type);
C
chengduo 已提交
400 401
  asserts_.emplace_back([=](Node *x) {
    for (auto *op : x->outputs) {
402 403 404
      if (op->IsOp() && op->Op()->Type() == op_type &&
          IsNthInput(x, op, argument, nth))
        return true;
Y
Yan Chunwei 已提交
405 406 407 408 409
    }
    return false;
  });
  return this;
}
C
chengduo 已提交
410 411 412

PDNode *PDNode::assert_is_op_nth_output(const std::string &op_type,
                                        const std::string &argument, int nth) {
Y
Yan Chunwei 已提交
413
  assert_is_var();
C
chengduo 已提交
414 415
  asserts_.emplace_back([=](Node *x) {
    for (auto *op : x->inputs) {
416 417 418
      if (op->IsOp() && op->Op()->Type() == op_type &&
          IsNthOutput(x, op, argument, nth))
        return true;
Y
Yan Chunwei 已提交
419 420 421 422 423
    }
    return false;
  });
  return this;
}
C
chengduo 已提交
424 425

PDNode *PDNode::assert_is_only_input_of_op(const std::string &op_type) {
Y
Yan Chunwei 已提交
426
  assert_is_var();
C
chengduo 已提交
427 428
  asserts_.emplace_back([=](Node *x) {
    for (auto *op : x->outputs) {
Y
Yan Chunwei 已提交
429 430 431 432 433 434 435 436 437
      if (op && op->IsOp() && op->Op() && op->Op()->Type() == op_type &&
          op->inputs.size() == 1) {
        return true;
      }
    }
    return false;
  });
  return this;
}
C
chengduo 已提交
438 439

PDNode *PDNode::assert_is_only_output_of_op(const std::string &op_type) {
Y
Yan Chunwei 已提交
440
  assert_is_var();
C
chengduo 已提交
441 442
  asserts_.emplace_back([=](Node *x) {
    for (auto *op : x->inputs) {
Y
Yan Chunwei 已提交
443 444 445 446 447 448 449 450 451
      if (op && op->IsOp() && op->Op() && op->Op()->Type() == op_type &&
          op->outputs.size() == 1) {
        return true;
      }
    }
    return false;
  });
  return this;
}
C
chengduo 已提交
452 453

PDNode *PDNode::assert_is_op_output(const std::string &op_type) {
Y
Yan Chunwei 已提交
454
  assert_is_var();
C
chengduo 已提交
455 456
  asserts_.emplace_back([=](Node *x) {
    for (auto *op : x->inputs) {
Y
Yan Chunwei 已提交
457 458 459 460 461 462 463 464
      if (op && op->IsOp() && op->Op() && op->Op()->Type() == op_type) {
        return true;
      }
    }
    return false;
  });
  return this;
}
C
chengduo 已提交
465 466 467

PDNode *PDNode::assert_is_op_output(const std::string &op_type,
                                    const std::string &argument) {
468 469 470 471
  assert_is_var();
  assert_is_op_nth_output(op_type, argument, 0);
  return this;
}
C
chengduo 已提交
472
PDNode *PDNode::assert_is_op_input(const std::string &op_type) {
Y
Yan Chunwei 已提交
473
  assert_is_var();
C
chengduo 已提交
474 475
  asserts_.emplace_back([=](Node *x) {
    for (auto *op : x->outputs) {
Y
Yan Chunwei 已提交
476 477 478 479 480 481 482 483
      if (op && op->IsOp() && op->Op() && op->Op()->Type() == op_type) {
        return true;
      }
    }
    return false;
  });
  return this;
}
C
chengduo 已提交
484 485 486

PDNode *PDNode::assert_is_op_input(const std::string &op_type,
                                   const std::string &argument) {
487 488 489 490
  assert_is_var();
  assert_is_op_nth_input(op_type, argument, 0);
  return this;
}
C
chengduo 已提交
491 492

PDNode *PDNode::assert_op_has_n_inputs(const std::string &op_type, size_t n) {
Y
Yan Chunwei 已提交
493
  assert_is_op(op_type);
C
chengduo 已提交
494
  asserts_.emplace_back([=](Node *x) { return x->inputs.size() == n; });
Y
Yan Chunwei 已提交
495 496
  return this;
}
C
chengduo 已提交
497 498

PDNode *PDNode::assert_op_has_n_outputs(const std::string &op_type, size_t n) {
Y
Yan Chunwei 已提交
499
  assert_is_op(op_type);
C
chengduo 已提交
500
  asserts_.emplace_back([=](Node *x) { return x->outputs.size() == n; });
Y
Yan Chunwei 已提交
501 502
  return this;
}
C
chengduo 已提交
503 504

PDNode *PDNode::assert_more(PDNode::teller_t &&teller) {
Y
Yan Chunwei 已提交
505 506 507 508
  asserts_.emplace_back(std::move(teller));
  return this;
}

C
chengduo 已提交
509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591
PDNode *PDNode::assert_is_ops(const std::unordered_set<std::string> &op_types) {
  asserts_.emplace_back([op_types](Node *x) {
    return x && x->IsOp() && op_types.count(x->Op()->Type());
  });
  return this;
}

PDNode *PDNode::assert_is_ops_nth_input(
    const std::unordered_set<std::string> &op_types,
    const std::string &argument, int nth) {
  assert_is_var();
  assert_is_ops_input(op_types);
  asserts_.emplace_back([=](Node *x) {
    for (auto *op : x->outputs) {
      if (op->IsOp() && op_types.count(op->Op()->Type()) &&
          IsNthInput(x, op, argument, nth))
        return true;
    }
    return false;
  });
  return this;
}

PDNode *PDNode::assert_is_ops_nth_output(
    const std::unordered_set<std::string> &op_types,
    const std::string &argument, int nth) {
  assert_is_var();
  asserts_.emplace_back([=](Node *x) {
    for (auto *op : x->inputs) {
      if (op->IsOp() && op_types.count(op->Op()->Type()) &&
          IsNthOutput(x, op, argument, nth))
        return true;
    }
    return false;
  });
  return this;
}
PDNode *PDNode::assert_is_ops_output(
    const std::unordered_set<std::string> &op_types) {
  assert_is_var();
  asserts_.emplace_back([=](Node *x) {
    for (auto *op : x->inputs) {
      if (op && op->IsOp() && op->Op() && op_types.count(op->Op()->Type())) {
        return true;
      }
    }
    return false;
  });
  return this;
}

PDNode *PDNode::assert_is_ops_output(
    const std::unordered_set<std::string> &op_types,
    const std::string &argument) {
  assert_is_var();
  assert_is_ops_nth_output(op_types, argument, 0);
  return this;
}

PDNode *PDNode::assert_is_ops_input(
    const std::unordered_set<std::string> &op_types) {
  assert_is_var();
  asserts_.emplace_back([=](Node *x) {
    for (auto *op : x->outputs) {
      if (op && op->IsOp() && op->Op() && op_types.count(op->Op()->Type())) {
        return true;
      }
    }
    return false;
  });
  return this;
}

PDNode *PDNode::assert_is_ops_input(
    const std::unordered_set<std::string> &op_types,
    const std::string &argument) {
  assert_is_var();
  assert_is_ops_nth_input(op_types, argument, 0);
  return this;
}

bool VarLinksToOp(Node *node, const std::string &op_type) {
  for (auto *out : node->outputs) {
592 593 594 595 596 597
    if (out->IsOp() && out->Op()->Type() == op_type) {
      return true;
    }
  }
  return false;
}
C
chengduo 已提交
598 599

bool IsNthInput(Node *var, Node *op, const std::string &argument, size_t nth) {
600 601 602 603 604
  PADDLE_ENFORCE(var->IsVar());
  PADDLE_ENFORCE(op->IsOp());
  if (op->Op()->Input(argument).size() <= nth) return false;
  return var->Name() == op->Op()->Input(argument)[nth];
}
C
chengduo 已提交
605 606

bool IsNthOutput(Node *var, Node *op, const std::string &argument, size_t nth) {
607 608 609 610 611
  PADDLE_ENFORCE(var->IsVar());
  PADDLE_ENFORCE(op->IsOp());
  if (op->Op()->Output(argument).size() <= nth) return false;
  return var->Name() == op->Op()->Output(argument)[nth];
}
C
chengduo 已提交
612 613 614 615 616

void GraphSafeRemoveNodes(Graph *graph,
                          const std::unordered_set<const Node *> &nodes) {
  for (auto *node : nodes) {
    graph->RemoveNode(const_cast<Node *>(node));
617 618
  }

C
chengduo 已提交
619
  for (auto *node : graph->Nodes()) {
620 621
    for (auto it = node->inputs.begin(); it != node->inputs.end();) {
      if (nodes.count(*it)) {
C
chengduo 已提交
622
        it = const_cast<Node *>(node)->inputs.erase(it);
623
      } else {
624
        it++;
625
      }
626 627 628
    }
    for (auto it = node->outputs.begin(); it != node->outputs.end();) {
      if (nodes.count(*it)) {
C
chengduo 已提交
629
        it = const_cast<Node *>(node)->outputs.erase(it);
630
      } else {
631
        it++;
632
      }
633 634 635
    }
  }
}
C
chengduo 已提交
636 637 638

bool VarLinksFromOp(Node *node, const std::string &op_type) {
  for (auto *out : node->inputs) {
639 640 641 642 643 644 645
    if (out->IsOp() && out->Op()->Type() == op_type) {
      return true;
    }
  }
  return false;
}

S
Sylwester Fraczek 已提交
646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751
PDNode *patterns::ConvBN::operator()(paddle::framework::ir::PDNode *conv_input,
                                     bool with_eltwise_add) {
  // Create Operators
  conv_input->assert_is_op_input("conv2d", "Input");
  auto *conv_op = pattern->NewNode(conv_repr())->assert_is_op("conv2d");

  PDNode *eltwise_op = nullptr;
  if (with_eltwise_add) {
    eltwise_op =
        pattern->NewNode(eltwise_repr())->assert_is_op("elementwise_add");
  }
  auto *batch_norm_op =
      pattern->NewNode(batch_norm_repr())->assert_is_op("batch_norm");
  // Create variables
  // Conv Filter
  auto *conv_weight_var = pattern->NewNode(conv_weight_repr())
                              ->AsInput()
                              ->assert_is_persistable_var()
                              ->assert_is_op_input("conv2d", "Filter");

  auto *conv_out_var = pattern->NewNode(conv_out_repr())
                           ->AsIntermediate()
                           ->assert_is_only_output_of_op("conv2d");

  PDNode *eltwise_y_in_var = nullptr;
  PDNode *eltwise_out_var = nullptr;
  if (with_eltwise_add) {
    // Conv output as Bias input
    conv_out_var->assert_is_op_input("elementwise_add", "X");
    // Bias
    eltwise_y_in_var = pattern->NewNode(eltwise_y_in_repr())
                           ->assert_is_op_input("elementwise_add", "Y")
                           ->AsInput();
    eltwise_out_var = pattern->NewNode(eltwise_out_repr())
                          ->AsIntermediate()
                          ->assert_is_only_output_of_op("elementwise_add");
  } else {
    // Conv output as BN input
    conv_out_var->assert_is_op_input("batch_norm", "X");
  }

  // BN Scale
  auto *bn_scale_var = pattern->NewNode(bn_scale_repr())
                           ->AsInput()
                           ->assert_is_persistable_var()
                           ->assert_is_op_input("batch_norm", "Scale");
  // BN Bias
  auto *bn_bias_var = pattern->NewNode(bn_bias_repr())
                          ->AsInput()
                          ->assert_is_persistable_var()
                          ->assert_is_op_input("batch_norm", "Bias");
  // BN Mean
  auto *bn_mean_var = pattern->NewNode(bn_mean_repr())
                          ->AsInput()
                          ->assert_is_persistable_var()
                          ->assert_is_op_input("batch_norm", "Mean");
  // BN Variance
  auto *bn_variance_var = pattern->NewNode(bn_variance_repr())
                              ->AsInput()
                              ->assert_is_persistable_var()
                              ->assert_is_op_input("batch_norm", "Variance");

  // BN output
  auto *bn_out_var = pattern->NewNode(bn_out_repr())
                         ->AsOutput()
                         ->assert_is_op_output("batch_norm");

  auto *bn_mean_out_var = pattern->NewNode(bn_mean_out_repr())
                              ->AsOutput()
                              ->assert_is_op_output("batch_norm", "MeanOut");

  auto *bn_variance_out_var =
      pattern->NewNode(bn_variance_out_repr())
          ->AsOutput()
          ->assert_is_op_output("batch_norm", "VarianceOut");

  auto *bn_saved_mean_var =
      pattern->NewNode(bn_saved_mean_repr())
          ->AsOutput()
          ->assert_is_op_output("batch_norm", "SavedMean");

  auto *bn_saved_variance_var =
      pattern->NewNode(bn_saved_variance_repr())
          ->AsOutput()
          ->assert_is_op_output("batch_norm", "SavedVariance");

  conv_op->LinksFrom({conv_input, conv_weight_var}).LinksTo({conv_out_var});

  if (with_eltwise_add) {
    eltwise_op->LinksFrom({conv_out_var, eltwise_y_in_var})
        .LinksTo({eltwise_out_var});
    batch_norm_op
        ->LinksFrom({eltwise_out_var, bn_scale_var, bn_bias_var, bn_mean_var,
                     bn_variance_var})
        .LinksTo({bn_out_var, bn_mean_out_var, bn_variance_out_var,
                  bn_saved_mean_var, bn_saved_variance_var});
  } else {
    batch_norm_op
        ->LinksFrom({conv_out_var, bn_scale_var, bn_bias_var, bn_mean_var,
                     bn_variance_var})
        .LinksTo({bn_out_var, bn_mean_out_var, bn_variance_out_var,
                  bn_saved_mean_var, bn_saved_variance_var});
  }
  return bn_out_var;
}

C
chengduo 已提交
752 753
PDNode *patterns::ConvReLU::operator()(
    paddle::framework::ir::PDNode *conv_input) {
754 755
  // Create Operators
  conv_input->assert_is_op_input("conv2d", "Input");
C
chengduo 已提交
756 757
  auto *conv_op = pattern->NewNode(conv_repr())->assert_is_op("conv2d");
  auto *relu_op = pattern->NewNode(relu_repr())->assert_is_op("relu");
758 759
  // Create variables
  // Filter
C
chengduo 已提交
760
  auto *conv_weight_var = pattern->NewNode(conv_weight_repr())
761 762 763 764
                              ->AsInput()
                              ->assert_is_persistable_var()
                              ->assert_is_op_input("conv2d", "Filter");
  // intermediate variable, will be removed in the IR after fuse.
C
chengduo 已提交
765
  auto *conv_out_var = pattern->NewNode(conv_out_repr())
766 767 768 769
                           ->AsIntermediate()
                           ->assert_is_only_output_of_op("conv2d")
                           ->assert_is_op_input("relu");
  // output
C
chengduo 已提交
770
  auto *relu_out_var = pattern->NewNode(relu_out_repr())
771 772 773
                           ->AsOutput()
                           ->assert_is_op_output("relu");

774
  conv_op->LinksFrom({conv_input, conv_weight_var}).LinksTo({conv_out_var});
775 776 777 778
  relu_op->LinksFrom({conv_out_var}).LinksTo({relu_out_var});
  return relu_out_var;
}

T
tensor-tang 已提交
779 780 781 782
PDNode *patterns::SeqConvEltAddRelu::operator()(
    paddle::framework::ir::PDNode *seqconv_input) {
  // Create Operators
  seqconv_input->assert_is_op_input("sequence_conv", "X");
T
tensor-tang 已提交
783 784 785 786
  auto *seqconv_op = pattern->NewNode(seqconv_repr())
                         ->assert_is_op("sequence_conv")
                         ->assert_op_attr<bool>("paddingTrainable", false)
                         ->assert_op_attr<int>("contextStride", 1);
T
tensor-tang 已提交
787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823

  auto *eltadd_op =
      pattern->NewNode(eltadd_repr())->assert_is_op("elementwise_add");
  auto *relu_op = pattern->NewNode(relu_repr())->assert_is_op("relu");
  // Create variables
  // Filter
  auto *seqconv_weight_var =
      pattern->NewNode(seqconv_weight_repr())
          ->AsInput()
          ->assert_is_persistable_var()
          ->assert_is_op_input("sequence_conv", "Filter");
  // Bias
  auto *eltadd_bias_var = pattern->NewNode(eltadd_bias_repr())
                              ->AsInput()
                              ->assert_is_op_input("elementwise_add");
  // intermediate variable, will be removed in the IR after fuse.
  auto *seqconv_out_var = pattern->NewNode(seqconv_out_repr())
                              ->AsIntermediate()
                              ->assert_is_only_output_of_op("sequence_conv")
                              ->assert_is_op_input("elementwise_add");
  auto *eltadd_out_var = pattern->NewNode(eltadd_out_repr())
                             ->AsIntermediate()
                             ->assert_is_only_output_of_op("elementwise_add")
                             ->assert_is_only_input_of_op("relu");
  // output
  auto *relu_out_var = pattern->NewNode(relu_out_repr())
                           ->AsOutput()
                           ->assert_is_op_output("relu");

  seqconv_op->LinksFrom({seqconv_input, seqconv_weight_var})
      .LinksTo({seqconv_out_var});
  eltadd_op->LinksFrom({seqconv_out_var, eltadd_bias_var})
      .LinksTo({eltadd_out_var});
  relu_op->LinksFrom({eltadd_out_var}).LinksTo({relu_out_var});
  return relu_out_var;
}

C
chengduo 已提交
824
PDNode *patterns::FC::operator()(paddle::framework::ir::PDNode *x,
Y
Yan Chunwei 已提交
825 826 827
                                 bool with_bias) {
  // Create shared nodes.
  x->assert_is_op_input("mul", "X");
C
chengduo 已提交
828
  auto *mul = pattern->NewNode(mul_repr())->assert_is_op("mul");
Y
Yan Chunwei 已提交
829

C
chengduo 已提交
830
  auto *mul_w_var = pattern->NewNode(w_repr())
Y
Yan Chunwei 已提交
831 832 833 834
                        ->AsInput()
                        ->assert_is_persistable_var()
                        ->assert_is_op_input("mul", "Y");

C
chengduo 已提交
835
  auto *mul_out_var =
Y
Yan Chunwei 已提交
836 837 838 839 840 841 842 843 844 845
      pattern->NewNode(mul_out_repr())->assert_is_op_output("mul");

  if (!with_bias) {  // not with bias
    // Add links.
    mul->LinksFrom({x, mul_w_var}).LinksTo({mul_out_var});
    return mul_out_var;

  } else {  // with bias
    mul_out_var->AsIntermediate()->assert_is_op_input("elementwise_add");
    // Create operators.
C
chengduo 已提交
846
    auto *elementwise_add = pattern->NewNode(elementwise_add_repr())
Y
Yan Chunwei 已提交
847 848
                                ->assert_is_op("elementwise_add");
    // Create variables.
C
chengduo 已提交
849
    auto *bias = pattern->NewNode(bias_repr())
Y
Yan Chunwei 已提交
850 851 852
                     ->assert_is_op_input("elementwise_add")
                     ->AsInput();

C
chengduo 已提交
853
    auto *fc_out = pattern->NewNode(Out_repr())
Y
Yan Chunwei 已提交
854 855 856 857 858 859
                       ->AsOutput()
                       ->assert_is_op_output("elementwise_add");

    mul->LinksFrom({mul_w_var, x}).LinksTo({mul_out_var});
    elementwise_add->LinksFrom({mul_out_var, bias}).LinksTo({fc_out});
    return fc_out;
860 861
  }
}
T
tensor-tang 已提交
862

863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880
PDNode *patterns::Embedding::operator()(PDNode *x) {
  x->assert_is_op_input("lookup_table", "Ids");
  auto *lookup_table_op =
      pattern->NewNode(lookup_table_repr())->assert_is_op("lookup_table");
#define NEW_NODE(arg__, io__)                    \
  auto *arg__ = pattern->NewNode(arg__##_repr()) \
                    ->assert_is_op_##io__("lookup_table", #arg__);

  NEW_NODE(W, input);

  NEW_NODE(Out, output);
#undef NEW_NODE

  lookup_table_op->LinksFrom({x, W});
  lookup_table_op->LinksTo({Out});
  return Out;
}

C
chengduo 已提交
881
PDNode *patterns::LSTM::operator()(PDNode *x) {
882
  x->assert_is_op_input("lstm", "Input");
C
chengduo 已提交
883
  auto *lstm_op = pattern->NewNode(lstm_repr())->assert_is_op("lstm");
Y
Yan Chunwei 已提交
884
#define NEW_NODE(arg__, io__) \
C
chengduo 已提交
885
  auto *arg__ =               \
Y
Yan Chunwei 已提交
886
      pattern->NewNode(arg__##_repr())->assert_is_op_##io__("lstm", #arg__);
887 888 889 890 891

  // Currently, the H0 and C0 are optional
  // TODO(Superjomn) upgrade the fuse framework to support optional.
  // NEW_NODE(H0, input);
  // NEW_NODE(C0, input);
Y
Yan Chunwei 已提交
892 893
  NEW_NODE(Weight, input);
  NEW_NODE(Bias, input);
894

Y
Yan Chunwei 已提交
895 896 897 898 899
  NEW_NODE(Hidden, output);
  NEW_NODE(Cell, output);
  NEW_NODE(BatchGate, output);
  NEW_NODE(BatchCellPreAct, output);
#undef NEW_NODE
900 901 902 903 904

  lstm_op->LinksFrom({x, Weight, Bias});
  lstm_op->LinksTo({Hidden, Cell, BatchGate, BatchCellPreAct});
  return Hidden;
}
T
tensor-tang 已提交
905

C
chengduo 已提交
906
PDNode *patterns::GRU::operator()(PDNode *x) {
T
tensor-tang 已提交
907
  x->assert_is_op_input("gru", "Input");
C
chengduo 已提交
908
  auto *gru_op = pattern->NewNode(gru_repr())->assert_is_op("gru");
Y
Yan Chunwei 已提交
909
#define NEW_NODE(arg__, io__) \
C
chengduo 已提交
910
  auto *arg__ =               \
Y
Yan Chunwei 已提交
911
      pattern->NewNode(arg__##_repr())->assert_is_op_##io__("gru", #arg__);
T
tensor-tang 已提交
912

Y
Yan Chunwei 已提交
913
  NEW_NODE(Weight, input);
T
tensor-tang 已提交
914 915
  // TODO(Superjomn): upgrade the fuse framework to support optional.
  // H0 and bias are optional
Y
Yan Chunwei 已提交
916
  NEW_NODE(Bias, input);  // also optional
T
tensor-tang 已提交
917 918
  // NEW_NODE(H0, input);

Y
Yan Chunwei 已提交
919
  NEW_NODE(Hidden, output);
T
tensor-tang 已提交
920
  // below are intermediate
Y
Yan Chunwei 已提交
921 922 923 924
  NEW_NODE(BatchGate, output);
  NEW_NODE(BatchResetHiddenPrev, output);
  NEW_NODE(BatchHidden, output);
#undef NEW_NODE
T
tensor-tang 已提交
925

T
tensor-tang 已提交
926 927 928 929
  BatchGate->AsIntermediate();
  BatchResetHiddenPrev->AsIntermediate();
  BatchHidden->AsIntermediate();

T
tensor-tang 已提交
930 931 932 933 934
  gru_op->LinksFrom({x, Weight, Bias});
  gru_op->LinksTo({Hidden, BatchGate, BatchResetHiddenPrev, BatchHidden});
  return Hidden;
}

C
chengduo 已提交
935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028
PDNode *patterns::ActElewiseAdd::operator()(
    paddle::framework::ir::PDNode *in_var,
    std::unordered_set<std::string> act_types) {
  in_var->assert_is_ops_input(act_types, "X");

  auto *act = pattern->NewNode(act_repr())->assert_is_ops(act_types);
  auto *act_out_var = pattern->NewNode(act_out_repr())
                          ->assert_is_not_ctrl_var()
                          ->assert_is_ops_output(act_types);
  act_out_var->AsIntermediate()->assert_is_op_input("elementwise_add");

  auto *ele_x_var = pattern->NewNode(ele_x_repr())
                        ->assert_is_not_ctrl_var()
                        ->assert_is_op_input("elementwise_add")
                        ->AsInput();
  auto *elementwise_add =
      pattern->NewNode(ele_add_repr())->assert_is_op("elementwise_add");

  auto *elewise_add_out = pattern->NewNode(elewise_add_out_repr())
                              ->AsOutput()
                              ->assert_is_op_output("elementwise_add", "Out");

  act->LinksFrom({in_var}).LinksTo({act_out_var});
  elementwise_add->LinksFrom({act_out_var, ele_x_var})
      .LinksTo({elewise_add_out});

  return elewise_add_out;
}

PDNode *patterns::ElewiseAddAct::operator()(
    paddle::framework::ir::PDNode *ele_x_var,
    std::unordered_set<std::string> act_types) {
  auto *ele_y_var = pattern->NewNode(ele_y_repr())
                        ->assert_is_op_input("elementwise_add", "Y");

  auto *ele_add =
      pattern->NewNode(ele_add_repr())->assert_is_op("elementwise_add");

  auto *ele_out_var = pattern->NewNode(elewise_add_out_repr())
                          ->assert_is_op_output("elementwise_add", "Out");

  ele_out_var->AsIntermediate()->assert_is_ops_input(act_types);

  auto *act = pattern->NewNode(act_repr())->assert_is_ops(act_types);

  auto *act_out_var =
      pattern->NewNode(act_out_repr())->assert_is_ops_output(act_types, "Out");

  ele_add->LinksFrom({ele_x_var, ele_y_var}).LinksTo({ele_out_var});
  act->LinksFrom({ele_out_var}).LinksTo({act_out_var});

  return act_out_var;
}

PDNode *patterns::ElewiseAddActInplaceGrad::operator()(
    paddle::framework::ir::PDNode *d_act_out_var,
    std::unordered_set<std::string> act_types) {
  // act_grad: in["Out", "Out@GRAD"], out["X@GRAD"]
  // ele_add_grad: in["Y", "Out@GRAD"], out["X@GRAD", "Y@GRAD"]
  auto *act_grad = pattern->NewNode(act_grad_repr())->assert_is_ops(act_types);

  auto *act_out_var =
      pattern->NewNode(act_out_repr())->assert_is_ops_input(act_types, "Out");

  auto *d_intermediate_var =
      pattern->NewNode(d_itermediate_out_repr())
          ->assert_is_ops_output(act_types, GradVarName("X"));

  act_grad->LinksFrom({d_act_out_var, act_out_var})
      .LinksTo({d_intermediate_var});

  auto *ele_y_var = pattern->NewNode(ele_y_repr())
                        ->assert_is_not_ctrl_var()
                        ->assert_is_op_input("elementwise_add_grad", "Y");

  auto *ele_add_grad = pattern->NewNode(ele_add_grad_repr())
                           ->assert_is_op("elementwise_add_grad");

  auto *d_ele_x_var =
      pattern->NewNode(d_ele_x_repr())
          ->assert_is_not_ctrl_var()
          ->assert_is_op_output("elementwise_add_grad", GradVarName("X"));

  auto *d_ele_y_var =
      pattern->NewNode(d_ele_y_repr())
          ->assert_is_not_ctrl_var()
          ->assert_is_op_output("elementwise_add_grad", GradVarName("Y"));

  ele_add_grad->LinksFrom({d_intermediate_var, ele_y_var})
      .LinksTo({d_ele_x_var, d_ele_y_var});

  return ele_add_grad;
}

M
Michal Gallus 已提交
1029
PDNode *patterns::ConvBias::operator()(
1030
    paddle::framework::ir::PDNode *conv_input, bool is_conv3d) {
Y
Yihua Xu 已提交
1031
  std::string type = is_conv3d ? "conv3d" : "conv2d";
M
Michal Gallus 已提交
1032
  // Create Operators
Y
Yihua Xu 已提交
1033 1034
  conv_input->assert_is_op_input(type, "Input");
  auto *conv_op = pattern->NewNode(conv_repr())->assert_is_op(type);
M
Michal Gallus 已提交
1035 1036 1037 1038
  auto *eltiwse_op =
      pattern->NewNode(eltwise_repr())->assert_is_op("elementwise_add");
  // Create variables
  // Filter
Y
Yihua Xu 已提交
1039 1040 1041 1042
  auto *conv_weight_var = pattern->NewNode(conv_weight_repr())
                              ->AsInput()
                              ->assert_is_persistable_var()
                              ->assert_is_op_input(type, "Filter");
M
Michal Gallus 已提交
1043
  // intermediate variable, will be removed in the IR after fuse.
Y
Yihua Xu 已提交
1044 1045 1046 1047
  auto *conv_out_var = pattern->NewNode(conv_out_repr())
                           ->AsIntermediate()
                           ->assert_is_only_output_of_op(type)
                           ->assert_is_op_input("elementwise_add");
M
Michal Gallus 已提交
1048 1049 1050
  // Bias stored in elementwise_add
  auto *eltwise_bias_var = pattern->NewNode(eltwise_bias_repr())
                               ->AsInput()
M
Michal Gallus 已提交
1051
                               ->assert_is_persistable_var()
M
Michal Gallus 已提交
1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062
                               ->assert_is_op_input("elementwise_add", "Y");
  // output
  auto *eltwise_out_var = pattern->NewNode(eltwise_out_repr())
                              ->AsOutput()
                              ->assert_is_op_output("elementwise_add");
  conv_op->LinksFrom({conv_input, conv_weight_var}).LinksTo({conv_out_var});
  eltiwse_op->LinksFrom({conv_out_var, eltwise_bias_var})
      .LinksTo({eltwise_out_var});
  return eltwise_out_var;
}

1063 1064 1065 1066
PDNode *patterns::Conv::operator()() {
  auto conv_op = pattern->NewNode(conv_op_repr())->assert_is_op("conv2d");

  auto input_var = pattern->NewNode(conv_input_repr())
1067
                       ->AsInput()
1068 1069 1070
                       ->assert_is_op_input("conv2d", "Input");

  auto filter_var = pattern->NewNode(conv_filter_repr())
1071
                        ->AsInput()
1072 1073 1074
                        ->assert_is_op_input("conv2d", "Filter");

  auto output_var = pattern->NewNode(conv_output_repr())
1075
                        ->AsOutput()
1076 1077
                        ->assert_is_op_output("conv2d", "Output");

1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 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
  conv_op->LinksFrom({input_var, filter_var}).LinksTo({output_var});
  return output_var;
}

PDNode *patterns::ConvResidual::operator()(bool with_residual_data) {
  auto conv_op = pattern->NewNode(conv_op_repr())->assert_is_op("conv2d");

  if (!with_residual_data)
    conv_op->assert_op_attr("fuse_residual_connection", false);

  auto input_var = pattern->NewNode(conv_input_repr())
                       ->AsInput()
                       ->assert_is_op_input("conv2d", "Input");

  auto filter_var = pattern->NewNode(conv_filter_repr())
                        ->AsInput()
                        ->assert_is_op_input("conv2d", "Filter");

  auto output_var = pattern->NewNode(conv_output_repr())
                        ->AsOutput()
                        ->assert_is_op_output("conv2d", "Output");

  std::vector<PDNode *> links_from{input_var, filter_var};

  if (with_residual_data) {
    auto res_conn_var = pattern->NewNode(conv_residual_data_repr())
                            ->AsInput()
                            ->assert_is_op_input("conv2d", "ResidualData");
    links_from.push_back(res_conn_var);
  }

  conv_op->LinksFrom(links_from).LinksTo({output_var});
  return output_var;
}

PDNode *patterns::Pool::operator()() {
  auto pool_op = pattern->NewNode(pool_op_repr())->assert_is_op("pool2d");

  auto input_var = pattern->NewNode(pool_input_repr())
                       ->AsInput()
                       ->assert_is_op_input("pool2d", "X");

  auto output_var = pattern->NewNode(pool_output_repr())
                        ->AsOutput()
                        ->assert_is_op_output("pool2d", "Out");
1123

1124
  pool_op->LinksFrom({input_var}).LinksTo({output_var});
1125 1126 1127
  return output_var;
}

1128
PDNode *patterns::ElementwiseAdd::operator()(PDNode *x_var, PDNode *y_var) {
1129 1130 1131
  auto elementwise_add_op = pattern->NewNode(elementwise_add_op_repr())
                                ->assert_is_op("elementwise_add");

1132 1133
  x_var->AsInput()->assert_is_op_input("elementwise_add", "X");
  y_var->AsInput()->assert_is_op_input("elementwise_add", "Y");
1134 1135 1136 1137
  auto out_var = pattern->NewNode(elementwise_add_out_repr())
                     ->AsOutput()
                     ->assert_is_op_output("elementwise_add", "Out");

1138
  elementwise_add_op->LinksFrom({x_var, y_var});
1139 1140 1141 1142
  elementwise_add_op->LinksTo({out_var});

  return out_var;
}
1143

H
hjchen2 已提交
1144
std::unordered_set<std::string> conv_act_set({"identity", "relu"});
1145 1146 1147 1148 1149 1150 1151 1152 1153 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 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209

PDNode *patterns::ConvElementwiseaddAct::operator()(PDNode *conv_in) {
  conv_in->AsInput();
  auto conv_op = pattern->NewNode(conv_op_repr())->assert_is_op("conv2d");
  auto conv_out = pattern->NewNode(conv_out_repr())
                      ->assert_is_op_output("conv2d")
                      ->assert_is_op_input("elementwise_add", "X")
                      ->AsIntermediate();
  auto conv_filter = pattern->NewNode(conv_filter_repr())
                         ->assert_is_op_input("conv2d", "Filter")
                         ->AsInput();
  auto elementwise_add_op = pattern->NewNode(elementwise_add_op_repr())
                                ->assert_is_op("elementwise_add");
  auto elementwise_add_in_y = pattern->NewNode(elementwise_add_in_y_repr())
                                  ->assert_is_op_input("elementwise_add", "Y")
                                  ->AsInput();
  auto elementwise_add_out = pattern->NewNode(elementwise_add_out_repr())
                                 ->assert_is_op_output("elementwise_add")
                                 ->AsIntermediate();

  auto act_op = pattern->NewNode(act_op_repr())
                    ->assert_is_op()
                    ->assert_more([&](Node *node) {
                      auto op_type = node->Name();
                      return conv_act_set.count(op_type);
                    });

  auto act_out = pattern->NewNode(act_out_repr())
                     ->assert_is_var()
                     // is activation op's output.
                     ->assert_more([&](Node *node) {
                       for (auto *in_op : node->inputs) {
                         if (conv_act_set.count(in_op->Name())) {
                           return true;
                         }
                       }
                       return false;
                     })
                     ->AsOutput();

  conv_op->LinksFrom({conv_in, conv_filter});
  conv_out->LinksFrom({conv_op});
  elementwise_add_op->LinksFrom({conv_out, elementwise_add_in_y})
      .LinksTo({elementwise_add_out});
  act_op->LinksFrom({elementwise_add_out}).LinksTo({act_out});

  return act_out;
}

PDNode *patterns::ConvElementwiseadd2Act::operator()(PDNode *conv_in) {
  auto conv_op = pattern->NewNode(conv_op_repr())->assert_is_op("conv2d");
  auto conv_filter = pattern->NewNode(conv_filter_repr())
                         ->assert_is_op_input("conv2d", "Filter")
                         ->AsInput();
  auto conv_out = pattern->NewNode(conv_out_repr())
                      ->assert_is_op_output("conv2d")
                      ->assert_is_op_input("elementwise_add", "X")
                      ->AsIntermediate();
  auto elementwise_add_op = pattern->NewNode(elementwise_add_op_repr())
                                ->assert_is_op("elementwise_add");
  auto elementwise_add_in_y = pattern->NewNode(elementwise_add_in_y_repr())
                                  ->assert_is_op_input("elementwise_add", "Y")
                                  ->AsInput();
  auto elementwise_add_out = pattern->NewNode(elementwise_add_out_repr())
                                 ->assert_is_op_output("elementwise_add")
H
hjchen2 已提交
1210
                                 ->assert_is_op_input("elementwise_add", "Y")
1211 1212 1213 1214 1215
                                 ->AsIntermediate();

  auto elementwise_add_op_1 = pattern->NewNode(elementwise_add_op_1_repr())
                                  ->assert_is_op("elementwise_add");
  auto elementwise_add_in_y_1 = pattern->NewNode(elementwise_add_in_y_1_repr())
H
hjchen2 已提交
1216
                                    ->assert_is_op_input("elementwise_add", "X")
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
                                    ->AsInput();
  auto elementwise_add_out_1 = pattern->NewNode(elementwise_add_out_1_repr())
                                   ->assert_is_op_output("elementwise_add")
                                   ->AsIntermediate();

  auto act_op = pattern->NewNode(act_op_repr())
                    ->assert_is_op()
                    ->assert_more([&](Node *node) {
                      auto op_type = node->Name();
                      return conv_act_set.count(op_type);
                    });
  auto act_out = pattern->NewNode(act_out_repr())
                     ->assert_is_var()
                     // is activation op's output.
                     ->assert_more([&](Node *node) {
                       for (auto *in_op : node->inputs) {
                         if (conv_act_set.count(in_op->Name())) {
                           return true;
                         }
                       }
                       return false;
                     })
                     ->AsOutput();

  conv_op->LinksFrom({conv_in, conv_filter}).LinksTo({conv_out});
  elementwise_add_op->LinksFrom({conv_out, elementwise_add_in_y})
      .LinksTo({elementwise_add_out});
H
hjchen2 已提交
1244 1245
  elementwise_add_op_1->LinksFrom({elementwise_add_out, elementwise_add_in_y_1})
      .LinksTo({elementwise_add_out_1});
1246 1247 1248 1249
  act_op->LinksFrom({elementwise_add_out_1}).LinksTo({act_out});
  return act_out;
}

N
nhzlx 已提交
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 1275 1276
PDNode *patterns::ConvElementwiseadd::operator()(PDNode *conv_in) {
  conv_in->AsInput();
  auto conv_op = pattern->NewNode(conv_op_repr())->assert_is_op("conv2d");
  auto conv_out = pattern->NewNode(conv_out_repr())
                      ->assert_is_op_output("conv2d")
                      ->assert_is_op_input("elementwise_add", "X")
                      ->AsIntermediate();
  auto conv_filter = pattern->NewNode(conv_filter_repr())
                         ->assert_is_op_input("conv2d", "Filter")
                         ->AsInput();
  auto elementwise_add_op = pattern->NewNode(elementwise_add_op_repr())
                                ->assert_is_op("elementwise_add");
  auto elementwise_add_in_y = pattern->NewNode(elementwise_add_in_y_repr())
                                  ->assert_is_op_input("elementwise_add", "Y")
                                  ->AsInput();
  auto elementwise_add_out = pattern->NewNode(elementwise_add_out_repr())
                                 ->assert_is_op_output("elementwise_add")
                                 ->AsOutput();

  conv_op->LinksFrom({conv_in, conv_filter});
  conv_out->LinksFrom({conv_op});
  elementwise_add_op->LinksFrom({conv_out, elementwise_add_in_y})
      .LinksTo({elementwise_add_out});

  return elementwise_add_out;
}

N
nhzlx 已提交
1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348
PDNode *patterns::ConvAffineChannel::operator()(
    paddle::framework::ir::PDNode *conv_input, bool with_eltwise_add) {
  // Create Operators
  conv_input->assert_is_op_input("conv2d", "Input");
  auto *conv_op = pattern->NewNode(conv_repr())->assert_is_op("conv2d");

  PDNode *eltwise_op = nullptr;
  if (with_eltwise_add) {
    eltwise_op =
        pattern->NewNode(eltwise_repr())->assert_is_op("elementwise_add");
  }

  auto *affine_channel_op =
      pattern->NewNode(affine_channel_repr())->assert_is_op("affine_channel");
  // Create variables
  // Conv Filter
  auto *conv_weight_var = pattern->NewNode(conv_weight_repr())
                              ->AsInput()
                              ->assert_is_persistable_var()
                              ->assert_is_op_input("conv2d", "Filter");

  auto *conv_out_var = pattern->NewNode(conv_out_repr())
                           ->AsIntermediate()
                           ->assert_is_only_output_of_op("conv2d");

  PDNode *eltwise_y_in_var = nullptr;
  PDNode *eltwise_out_var = nullptr;
  if (with_eltwise_add) {
    // Conv output as Bias input
    conv_out_var->assert_is_op_input("elementwise_add", "X");
    // Bias
    eltwise_y_in_var = pattern->NewNode(eltwise_y_in_repr())
                           ->assert_is_op_input("elementwise_add", "Y")
                           ->AsInput();
    eltwise_out_var = pattern->NewNode(eltwise_out_repr())
                          ->AsIntermediate()
                          ->assert_is_only_output_of_op("elementwise_add");
  } else {
    // Conv output as AffineChannel input
    conv_out_var->assert_is_op_input("affine_channel", "X");
  }

  // AC Scale
  auto *ac_scale_var = pattern->NewNode(ac_scale_repr())
                           ->AsInput()
                           ->assert_is_persistable_var()
                           ->assert_is_op_input("affine_channel", "Scale");
  // AC Bias
  auto *ac_bias_var = pattern->NewNode(ac_bias_repr())
                          ->AsInput()
                          ->assert_is_persistable_var()
                          ->assert_is_op_input("affine_channel", "Bias");

  // AC output
  auto *ac_out_var = pattern->NewNode(ac_out_repr())
                         ->AsOutput()
                         ->assert_is_op_output("affine_channel");

  conv_op->LinksFrom({conv_input, conv_weight_var}).LinksTo({conv_out_var});

  if (with_eltwise_add) {
    eltwise_op->LinksFrom({conv_out_var, eltwise_y_in_var})
        .LinksTo({eltwise_out_var});
    affine_channel_op->LinksFrom({eltwise_out_var, ac_scale_var, ac_bias_var})
        .LinksTo({ac_out_var});
  } else {
    affine_channel_op->LinksFrom({conv_out_var, ac_scale_var, ac_bias_var})
        .LinksTo({ac_out_var});
  }
  return ac_out_var;
}

1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 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
PDNode *patterns::DequantQuantAny::operator()() {
  auto *dequant_in = pattern->NewNode(dequant_in_repr())
                         ->AsInput()
                         ->assert_is_op_input("dequantize", "Input");

  auto *dequant_op =
      pattern->NewNode(dequant_op_repr())->assert_is_op("dequantize");

  auto *dequant_out = pattern->NewNode(dequant_out_repr())
                          ->AsOutput()
                          ->assert_is_op_output("dequantize", "Output");

  auto *quant_op = pattern->NewNode(quant_op_repr())
                       ->assert_is_op("quantize")
                       ->AsIntermediate();

  auto *quant_out = pattern->NewNode(quant_out_repr())
                        ->AsOutput()
                        ->assert_is_op_output("quantize");

  auto *next_op = pattern->NewNode(next_op_repr())->assert_is_op();

  dequant_op->LinksFrom({dequant_in}).LinksTo({dequant_out});
  quant_op->LinksFrom({dequant_out}).LinksTo({quant_out});
  next_op->LinksFrom({quant_out});

  return quant_out;
}

PDNode *patterns::DequantAny::operator()() {
  auto *dequant_op =
      pattern->NewNode(dequant_op_repr())->assert_is_op("dequantize");

  auto *dequant_out = pattern->NewNode(dequant_out_repr())
                          ->AsOutput()
                          ->assert_is_op_output("dequantize", "Output");

  auto *next_op = pattern->NewNode(next_op_repr())->assert_is_op();

  dequant_op->LinksTo({dequant_out});
  next_op->LinksFrom({dequant_out});

  return dequant_out;
}

1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456
// a -> transpose_op(1) -> transpose_out_a -> flatten_op(1) -> flatten_out_a
// b -> transpose_op(2) -> transpose_out_b -> flatten_op(2) -> flatten_out_b
// ...
// z -> transpose_op(n) -> transpose_out_z -> flatten_op(n) -> flatten_out_z
// flatten_out_a -> concat_op  flatten_out_b -> concat_op ... flatten_out_z ->
// concat_op
PDNode *patterns::TransposeFlattenConcat::operator()(
    std::vector<PDNode *> conv_in, int times) {
  // The times represents the repeat times of the
  // {trans, trans_out, flatten, flatten_out}
  const int kNumFields = 4;
  const int kTransOutOffset = 1;
  const int kFlattenOffset = 2;
  const int kFlattenOutOffset = 3;

  std::vector<PDNode *> nodes;

  for (int i = 0; i < times; i++) {
    nodes.push_back(
        pattern->NewNode(GetNodeName("transpose" + std::to_string(i)))
            ->assert_is_op("transpose2"));
    nodes.push_back(
        pattern->NewNode(GetNodeName("transpose_out" + std::to_string(i)))
            ->assert_is_op_output("transpose2")
            ->assert_is_op_input("flatten2", "X")
            ->AsIntermediate());
    nodes.push_back(pattern->NewNode(GetNodeName("flatten" + std::to_string(i)))
                        ->assert_is_op("flatten2"));

    nodes.push_back(
        pattern->NewNode(GetNodeName("flatten_out" + std::to_string(i)))
            ->assert_is_op_output("flatten2")
            ->assert_is_op_nth_input("concat", "X", i)
            ->AsIntermediate());
  }

  auto concat_op = pattern->NewNode(GetNodeName("concat"))
                       ->assert_is_op("concat")
                       ->assert_op_has_n_inputs("concat", times);
  auto concat_out = pattern->NewNode(GetNodeName("concat_out"))
                        ->assert_is_op_output("concat")
                        ->AsOutput();

  std::vector<PDNode *> flatten_outs;
  for (int i = 0; i < times; i++) {
    conv_in[i]->AsInput();
    // trans
    nodes[i * kNumFields]->LinksFrom({conv_in[i]});
    // trans_out
    nodes[i * kNumFields + kTransOutOffset]->LinksFrom({nodes[i * kNumFields]});
    // flatten
    nodes[i * kNumFields + kFlattenOffset]->LinksFrom(
        {nodes[i * kNumFields + kTransOutOffset]});
    // flatten_out
    nodes[i * kNumFields + kFlattenOutOffset]->LinksFrom(
        {nodes[i * kNumFields + kFlattenOffset]});
    flatten_outs.push_back(nodes[i * kNumFields + kFlattenOutOffset]);
  }

  concat_op->LinksFrom(flatten_outs).LinksTo({concat_out});
  return concat_out;
}

1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529
PDNode *patterns::AnakinDetectionPattern::operator()(
    std::vector<PDNode *> conv_in, int times) {
  // The times represents the repeat times of the
  // {prior_box, prior_box_loc_out, flatten, prior_box_var_out, reshape}
  const int kNumFields = 7;
  const int kPriorBoxLocOffset = 1;
  const int kReshape1Offset = 2;
  const int kReshape1OutOffset = 3;
  const int kPriorBoxVarOffset = 4;
  const int kReshape2Offset = 5;
  const int kReshape2OutOffset = 6;

  const int kBoxCoderThirdInputOffset = times;
  const int kMultiClassSecondInputNmsOffset = times + 1;

  std::vector<PDNode *> nodes;

  for (int i = 0; i < times; i++) {
    nodes.push_back(
        pattern->NewNode(GetNodeName("prior_box" + std::to_string(i)))
            ->assert_is_op("density_prior_box"));
    nodes.push_back(pattern->NewNode(GetNodeName("box_out" + std::to_string(i)))
                        ->assert_is_op_output("density_prior_box", "Boxes")
                        ->assert_is_op_input("reshape2", "X")
                        ->AsIntermediate());
    nodes.push_back(
        pattern->NewNode(GetNodeName("reshape1" + std::to_string(i)))
            ->assert_is_op("reshape2"));

    nodes.push_back(
        pattern->NewNode(GetNodeName("reshape1_out" + std::to_string(i)))
            ->assert_is_op_output("reshape2")
            ->assert_is_op_nth_input("concat", "X", i)
            ->AsIntermediate());

    nodes.push_back(
        pattern->NewNode(GetNodeName("box_var_out" + std::to_string(i)))
            ->assert_is_op_output("density_prior_box", "Variances")
            ->assert_is_op_input("reshape2", "X")
            ->AsIntermediate());
    nodes.push_back(
        pattern->NewNode(GetNodeName("reshape2" + std::to_string(i)))
            ->assert_is_op("reshape2"));

    nodes.push_back(
        pattern->NewNode(GetNodeName("reshape2_out" + std::to_string(i)))
            ->assert_is_op_output("reshape2")
            ->assert_is_op_nth_input("concat", "X", i)
            ->AsIntermediate());
  }

  auto concat_op1 = pattern->NewNode(GetNodeName("concat1"))
                        ->assert_is_op("concat")
                        ->assert_op_has_n_inputs("concat", times);
  auto concat_out1 = pattern->NewNode(GetNodeName("concat1_out"))
                         ->assert_is_op_output("concat")
                         ->AsIntermediate();

  auto concat_op2 = pattern->NewNode(GetNodeName("concat2"))
                        ->assert_is_op("concat")
                        ->assert_op_has_n_inputs("concat", times);
  auto concat_out2 = pattern->NewNode(GetNodeName("concat2_out"))
                         ->assert_is_op_output("concat")
                         ->AsIntermediate();

  auto box_coder_op = pattern->NewNode(GetNodeName("box_coder"))
                          ->assert_is_op("box_coder")
                          ->assert_op_has_n_inputs("box_coder", 3);

  auto box_coder_out = pattern->NewNode(GetNodeName("box_coder_out"))
                           ->assert_is_op_output("box_coder")
                           ->AsIntermediate();

1530 1531 1532 1533 1534 1535 1536 1537 1538 1539
  auto transpose_before_nms =
      pattern->NewNode(GetNodeName("transpose_before_nms"))
          ->assert_is_op("transpose2");

  auto transpose_before_nms_out =
      pattern->NewNode(GetNodeName("transpose_before_nms_out"))
          ->assert_is_op_output("transpose2")
          ->assert_is_op_input("multiclass_nms", "Scores")
          ->AsIntermediate();

1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589
  auto multiclass_nms_op = pattern->NewNode(GetNodeName("multiclass_nms"))
                               ->assert_is_op("multiclass_nms")
                               ->assert_op_has_n_inputs("multiclass_nms", 2);

  auto multiclass_nms_out = pattern->NewNode(GetNodeName("multiclass_nms_out"))
                                ->assert_is_op_output("multiclass_nms")
                                ->AsOutput();

  std::vector<PDNode *> reshape1_outs;
  std::vector<PDNode *> reshape2_outs;

  for (int i = 0; i < times; i++) {
    conv_in[i]->AsInput();
    // prior_box
    nodes[i * kNumFields]->LinksFrom({conv_in[i]});
    // prior_box box out
    nodes[i * kNumFields + kPriorBoxLocOffset]->LinksFrom(
        {nodes[i * kNumFields]});
    // reshape
    nodes[i * kNumFields + kReshape1Offset]->LinksFrom(
        {nodes[i * kNumFields + kPriorBoxLocOffset]});
    // reshape_out
    nodes[i * kNumFields + kReshape1OutOffset]->LinksFrom(
        {nodes[i * kNumFields + kReshape1Offset]});

    nodes[i * kNumFields + kPriorBoxVarOffset]->LinksFrom(
        {nodes[i * kNumFields]});
    // reshape
    nodes[i * kNumFields + kReshape2Offset]->LinksFrom(
        {nodes[i * kNumFields + kPriorBoxVarOffset]});
    // reshape_out
    nodes[i * kNumFields + kReshape2OutOffset]->LinksFrom(
        {nodes[i * kNumFields + kReshape2Offset]});

    reshape1_outs.push_back(nodes[i * kNumFields + kReshape1OutOffset]);
    reshape2_outs.push_back(nodes[i * kNumFields + kReshape2OutOffset]);
  }

  concat_op1->LinksFrom(reshape1_outs);
  concat_op2->LinksFrom(reshape2_outs);
  concat_out1->LinksFrom({concat_op1});
  concat_out2->LinksFrom({concat_op2});

  conv_in[kBoxCoderThirdInputOffset]->AsInput();
  conv_in[kMultiClassSecondInputNmsOffset]->AsInput();

  box_coder_op->LinksFrom(
      {concat_out1, concat_out2, conv_in[kBoxCoderThirdInputOffset]});
  box_coder_out->LinksFrom({box_coder_op});

1590 1591 1592 1593
  transpose_before_nms->LinksFrom({conv_in[kMultiClassSecondInputNmsOffset]});
  transpose_before_nms_out->LinksFrom({transpose_before_nms});

  multiclass_nms_op->LinksFrom({box_coder_out, transpose_before_nms_out})
1594 1595 1596 1597 1598
      .LinksTo({multiclass_nms_out});

  return multiclass_nms_out;
}

1599 1600 1601
}  // namespace ir
}  // namespace framework
}  // namespace paddle