Paddle/paddle/framework/backward.cc

/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve.

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/framework/backward.h"
#include "paddle/operators/net_op.h"

#include <deque>
#include <list>
#include <memory>
#include <unordered_set>

#include "paddle/framework/block_desc.h"
#include "paddle/framework/op_registry.h"
#include "paddle/operators/net_op.h"

namespace paddle {
namespace framework {

static std::unordered_set<std::string>* g_ctrl_flow_ops_ = nullptr;
// Control Flow operators's backward is significantly different from
// computational operators. Hack Code here.
// We should design a better way to backward CtrlFlowOps.
static std::unordered_set<std::string>& CtrlFlowOps() {
  if (g_ctrl_flow_ops_ == nullptr) {
    g_ctrl_flow_ops_ = new std::unordered_set<std::string>{
        "increment", "lod_rank_table", "less_than"};
  }
  return *g_ctrl_flow_ops_;
}

static inline std::unique_ptr<OperatorBase> CreateGradOp(
    const OperatorBase& op, const std::unordered_set<std::string>& no_grad_set,
    std::unordered_map<std::string, std::string>* grad_to_var) {
  OpDesc op_desc;
  op_desc.SetInputMap(op.Inputs());
  op_desc.SetOutputMap(op.Outputs());
  op_desc.SetType(op.Type());
  op_desc.SetAttrMap(op.Attrs());
  auto& info = OpInfoMap::Instance().Get(op.Type());
  auto grad_descs = info.GradOpMaker()(op_desc, no_grad_set, grad_to_var, {});
  std::vector<std::unique_ptr<OperatorBase>> grad_ops;
  grad_ops.reserve(grad_descs.size());
  std::transform(grad_descs.begin(), grad_descs.end(),
                 std::back_inserter(grad_ops),
                 [](const std::unique_ptr<OpDesc>& grad_desc) {
                   return OpRegistry::CreateOp(*grad_desc);
                 });
  PADDLE_ENFORCE(!grad_ops.empty());
  if (grad_ops.size() == 1) {
    return std::move(grad_ops[0]);
  } else {
    auto net_op = new operators::NetOp();
    for (auto& grad_op : grad_ops) {
      net_op->AppendOp(std::move(grad_op));
    }
    net_op->CompleteAddOp();
    return std::unique_ptr<OperatorBase>(net_op);
  }
}

template <typename Map, typename T>
static void ForEachVarName(const Map& names, T callback) {
  for (auto& name : names) {
    for (auto& n : name.second) {
      if (callback(n)) return;
    }
  }
}

// return whether all the names + suffixes in the set
static bool AllInSet(
    const std::map<std::string, std::vector<std::string>>& names,
    const std::string& suffix, const std::unordered_set<std::string>& set) {
  bool all_in_set = true;
  ForEachVarName(names, [&all_in_set, &set, &suffix](const std::string& n) {
    all_in_set = set.find(n + suffix) != set.end();
    return !all_in_set;
  });
  return all_in_set;
}

static std::unique_ptr<OperatorBase> NOP() {
  auto net_op = new operators::NetOp();
  net_op->SetType("@NOP@");
  net_op->CompleteAddOp();
  return std::unique_ptr<OperatorBase>(net_op);
}

//  Get backward operator from a forward operator, a recursive implementation.
//
//  no_grad_names the gradient variable names without gradient calculating.
//
//  uniq_id is a unique index used inside recursively calling
//  BackwardRecursive. use `uid = uniq_id++;` to get the unique index, and
//  pass `uniq_id` through recursive calling.
//
//  returns The backward operator. In a simple situation, it may be a simple
//  operator, in a complex situation, it maybe a NetOp.
//
//  See Backward.h for details
static std::unique_ptr<OperatorBase> BackwardRecursive(
    const OperatorBase& forwardOp,
    std::unordered_set<std::string>& no_grad_names,
    std::unordered_map<std::string, std::string>* grad_to_var,
    size_t& uniq_id) {
  //  If all input gradients of forwarding operator do not need to calculate,
  //  just return an NOP. Not return null ptr because NOP does not take
  //  too much time for calculation, but it is useful for simplifying logic.
  if (AllInSet(forwardOp.Inputs() /*names*/, kGradVarSuffix /*suffix*/,
               no_grad_names /*set*/)) {
    return NOP();
  }

  //  All output gradients of forwarding operator do not need to calculate.
  //  Then all input gradients cannot be computed at all, and we put them into
  //  `no_grad_names` set. Return an NOP.
  if (AllInSet(forwardOp.Outputs() /*names*/, kGradVarSuffix /*suffix*/,
               no_grad_names /*set*/)) {
    ForEachVarName(forwardOp.Inputs(),
                   [&no_grad_names](const std::string& name) -> bool {
                     no_grad_names.insert(GradVarName(name));
                     return false;
                   });
    return NOP();
  }

  // Returned gradient network
  auto net = std::unique_ptr<operators::NetOp>(new operators::NetOp());

  if (forwardOp.IsNetOp()) {
    // Because forwardOp is a net op, it can static_cast.
    auto& forwardNet = static_cast<const operators::NetOp&>(forwardOp);

    // Map from output gradient variable name to operator's indices in
    // backward net's ops_. That operator generates that variable.
    std::unordered_map<std::string, std::vector<size_t>> dup_output_ops;

    size_t local_op_id = 0;
    // reversely travel forwardNet and collect all duplicate outputs.
    for (auto it = forwardNet.ops_.rbegin(); it != forwardNet.ops_.rend();
         ++it, ++local_op_id) {
      auto& fwd = *it;
      auto bwd = BackwardRecursive(*fwd, no_grad_names, grad_to_var, uniq_id);
      ForEachVarName(bwd->Outputs(),
                     [&dup_output_ops, local_op_id](const std::string& out) {
                       dup_output_ops[out].emplace_back(local_op_id);
                       return false;
                     });
      net->AppendOp(std::move(bwd));
    }
    // Get unique ID for this method.
    auto uid = uniq_id++;
    // TODO(dzh): more comment
    // multiple operators which have the same output (y for example) may
    // overwrite the same y variable when backward, special operations are token
    // to handle this case. For each duplicate output, rename it to an alias
    // (original name with a offset), append an `add` op for its operator,
    // and finally sum all the alias variable to the final output variable y.
    using Pos = std::pair<size_t, std::unique_ptr<OperatorBase>>;
    std::list<Pos> insert_position;
    for (auto& dup_output_op : dup_output_ops) {
      const std::string& name = dup_output_op.first;
      // duplicate @Empty@ don't need to be added
      if (name == kEmptyVarName) continue;

      auto& dup_op = dup_output_op.second;
      // no duplicate output
      if (dup_op.size() == 1) continue;

      // process the duplicate outputs
      std::vector<std::string> dup_outputs;
      for (size_t i = 0; i < dup_op.size(); ++i) {
        // rename each duplicate output to an alias
        auto op_offset = dup_op[i];
        dup_outputs.push_back(name + "@RENAME@" + std::to_string(uid) + "@" +
                              std::to_string(i));
        net->ops_[op_offset]->Rename(name, dup_outputs.back());
      }
      // collect all the offset for each alias,
      // insert a sum operator to add all aliases to output
      insert_position.push_back(
          {dup_op.back(),
           OpRegistry::CreateOp("sum", {{"X", dup_outputs}}, {{"Out", {name}}},
                                AttributeMap{})});
    }

    // make sure the inserted `sum` ops follow the BFS order.
    insert_position.sort(
        [](const Pos& l, const Pos& r) { return l.first > r.first; });

    for (auto& pos : insert_position) {
      net->InsertOp(pos.first + 1, std::move(pos.second));
    }
  } else {
    std::unique_ptr<OperatorBase> grad_op(
        CreateGradOp(forwardOp, no_grad_names, grad_to_var));

    ForEachVarName(grad_op->Inputs(), [&no_grad_names, &net, &grad_op](
                                          const std::string& grad_input) {
      if (no_grad_names.count(grad_input)) {
        // +1 for \0
        std::string prefix = grad_input.substr(
            0, grad_input.size() - sizeof(kGradVarSuffix) / sizeof(char) + 1);
        grad_op->Rename(grad_input, prefix + kZeroVarSuffix);

        // If part of input gradient of that operator is not calculated, fill
        // zero variables to that input gradient.
        net->AppendOp(OpRegistry::CreateOp("fill_zeros_like", {{"X", {prefix}}},
                                           {{"Out", {grad_input}}},
                                           AttributeMap{}));
      }
      return false;
    });

    ForEachVarName(grad_op->Outputs(),
                   [&no_grad_names, &grad_op](const std::string& grad_output) {
                     if (no_grad_names.count(grad_output)) {
                       grad_op->Rename(grad_output, kEmptyVarName);
                     }
                     return false;
                   });

    if (net->ops_.empty()) {  // Current no aux op is added to network
      return grad_op;
    }
    net->AppendOp(std::move(grad_op));
  }
  net->SetType("@GENERATED_BACKWARD@");
  net->CompleteAddOp();
  return std::unique_ptr<OperatorBase>(
      static_cast<OperatorBase*>(net.release()));
}

// See header for comments
std::unique_ptr<OperatorBase> Backward(
    const OperatorBase& forwardOp,
    const std::unordered_set<std::string>& no_grad_vars) {
  std::unordered_set<std::string> no_grad_names;
  no_grad_names.reserve(no_grad_vars.size() + 1);

  no_grad_names.insert(std::string(kEmptyVarName) + kGradVarSuffix);

  for (auto& name : no_grad_vars) {
    no_grad_names.insert(name + kGradVarSuffix);
  }
  size_t uid = 0;
  std::unordered_map<std::string, std::string> grad_to_var;
  return BackwardRecursive(forwardOp, no_grad_names, &grad_to_var, uid);
}

// ====================================  //

static bool AllGradInSet(const std::vector<std::string>& names,
                         const std::unordered_set<std::string>& set) {
  for (const std::string& name : names) {
    if (!set.count(GradVarName(name))) {
      return false;
    }
  }
  if (VLOG_IS_ON(10)) {
    std::ostringstream sout;
    sout << "All input {";
    for (auto& name : names) {
      sout << name << ",";
    }
    sout << "} is in {";
    for (auto& name : set) {
      sout << name << ",";
    }
    sout << "}";
    VLOG(10) << sout.str();
  }
  return true;
}

static std::string FwdName(const std::string& grad_name) {
  auto pos = grad_name.find("@GRAD");
  if (pos == std::string::npos) {
    return "";
  } else {
    return grad_name.substr(0, pos);
  }
}

static void CreateGradVarInBlock(
    size_t grad_op_start_index,
    const std::unordered_map<std::string, std::string>& param_name_map,
    BlockDesc* block_desc,
    std::unordered_map<std::string, GradVarInfo>* grad_var_record) {
  auto ops = block_desc->AllOps();
  for (size_t op_index = grad_op_start_index; op_index < ops.size();
       ++op_index) {
    std::unordered_set<std::string> new_vars;
    auto& ctrl_flow_ops = CtrlFlowOps();
    ForEachVarName(ops[op_index]->Outputs(),
                   [&](const std::string& grad_var_name) {
                     if (ctrl_flow_ops.find(ops[op_index]->Type()) !=
                         ctrl_flow_ops.end()) {
                       if (block_desc->HasVarRecursive(grad_var_name)) {
                         return false;
                       }
                     } else {
                       if (block_desc->HasVar(grad_var_name)) {
                         return false;
                       }
                     }
                     if (grad_var_name == framework::kEmptyVarName) {
                       return false;
                     }
                     auto var = block_desc->Var(grad_var_name);
                     VLOG(10) << "Creating Variable " << grad_var_name;
                     new_vars.insert(var->Name());
                     auto it = param_name_map.find(grad_var_name);
                     if (it == param_name_map.end()) {
                       return false;
                     }
                     auto param_var_name = it->second;
                     auto& grad_record = (*grad_var_record)[param_var_name];
                     grad_record.name_ = grad_var_name;
                     grad_record.block_idx_ = block_desc->ID();
                     grad_record.op_idx_ = static_cast<int>(op_index);
                     return false; /* not break */
                   });
    ops[op_index]->InferVarType(block_desc);
    for (auto& arg : ops[op_index]->OutputArgumentNames()) {
      if (new_vars.find(arg) == new_vars.end()) {
        continue;
      }
      auto pname = FwdName(arg);
      auto* param = block_desc->FindVarRecursive(pname);
      auto* grad = block_desc->FindVar(arg);
      if (param == nullptr) {
        grad->SetDataType(proto::DataType::FP32);
      } else {
        grad->SetDataType(param->GetDataType());
      }
    }
    ops[op_index]->InferShape(*block_desc);
  }
}

std::vector<std::unique_ptr<OpDesc>> MakeOpGrad(
    const OpDesc* op_desc, std::unordered_set<std::string>* no_grad_vars,
    std::unordered_map<std::string, std::string>* grad_to_var,
    const std::vector<BlockDesc*>& grad_block = std::vector<BlockDesc*>()) {
  std::vector<std::unique_ptr<OpDesc>> grad_op_descs;
  // All input gradients of forwarding operator do not need to calculate.
  const std::vector<std::string>& inputs = op_desc->InputArgumentNames();
  if (AllGradInSet(inputs, *no_grad_vars)) {
    VLOG(10) << "Drop operator  " << op_desc->Type();
    return grad_op_descs;  // empty vector
  }

  // All output gradients of forwarding operator do not need to calculate.
  const std::vector<std::string>& outputs = op_desc->OutputArgumentNames();

  if (AllGradInSet(outputs, *no_grad_vars)) {
    VLOG(10) << "Drop operator " << op_desc->Type();
    // FIXME: Hack code here
    auto& ctrl_flow_ops = CtrlFlowOps();
    if (ctrl_flow_ops.find(op_desc->Type()) == ctrl_flow_ops.end()) {
      // Only computational op need drop input's gradient.
      for (const std::string& name : inputs) {
        no_grad_vars->insert(GradVarName(name));
        VLOG(10) << " Also drop " << GradVarName(name);
      }
    }

    return grad_op_descs;  // empty vector
  }

  grad_op_descs =
      OpInfoMap::Instance()
          .Get(op_desc->Type())
          .GradOpMaker()(*op_desc, *no_grad_vars, grad_to_var, grad_block);

  std::list<std::unique_ptr<OpDesc>> pending_fill_zeros_ops;
  for (auto& desc : grad_op_descs) {
    for (const std::string& in_name : desc->InputArgumentNames()) {
      if (no_grad_vars->count(in_name)) {
        std::string prefix = in_name.substr(
            0, in_name.size() - sizeof(kGradVarSuffix) / sizeof(char) + 1);
        std::string new_name = prefix + kZeroVarSuffix;
        desc->Rename(in_name, new_name);
        std::unique_ptr<OpDesc> fill_zeros_op(
            new OpDesc("fill_zeros_like", {{"X", {prefix}}},
                       {{"Out", {new_name}}}, AttributeMap{}));
        pending_fill_zeros_ops.push_back(std::move(fill_zeros_op));
      }
    }
  }

  for (auto& p : pending_fill_zeros_ops) {
    grad_op_descs.insert(grad_op_descs.begin(), std::move(p));
  }
  return grad_op_descs;
}

static BlockDesc* CreateStepBlock(
    ProgramDesc& program_desc, std::unordered_set<std::string>* no_grad_vars,
    std::unordered_map<std::string, std::string>* grad_to_var,
    int step_block_idx);

std::vector<std::unique_ptr<OpDesc>> MakeBlockBackward(
    ProgramDesc& program_desc, int block_idx,
    std::unordered_set<std::string>* no_grad_vars,
    std::unordered_map<std::string, std::string>* grad_to_var) {
  VLOG(5) << "MakeBlockBackward";
  BlockDesc* cur_block = program_desc.MutableBlock(block_idx);
  std::vector<OpDesc*> op_descs = cur_block->AllOps();
  std::unordered_map<std::string, std::vector<size_t>> dup_out_ops;
  size_t grad_desc_idx = 0;
  std::vector<std::unique_ptr<OpDesc>> backward_descs;

  for (auto it = op_descs.rbegin(); it != op_descs.rend(); ++it) {
    VLOG(5) << "Making backward " << (*it)->Type() << " op";
    std::vector<std::unique_ptr<OpDesc>> op_grads;

    if ((*it)->Type() == "recurrent" || (*it)->Type() == "while" ||
        (*it)->Type() == "parallel_do") {
      int step_block_idx = (*it)->GetBlockAttr("sub_block");
      BlockDesc* backward_block = CreateStepBlock(program_desc, no_grad_vars,
                                                  grad_to_var, step_block_idx);
      op_grads = MakeOpGrad(*it, no_grad_vars, grad_to_var, {backward_block});
    } else if ((*it)->Type() == "conditional_block") {
      BlockDesc* backward_block =
          CreateStepBlock(program_desc, no_grad_vars, grad_to_var,
                          (*it)->GetBlockAttr("sub_block"));
      op_grads = MakeOpGrad(*it, no_grad_vars, grad_to_var, {backward_block});
    } else {
      op_grads = MakeOpGrad(*it, no_grad_vars, grad_to_var);
    }

    if (VLOG_IS_ON(10)) {
      std::ostringstream sout;
      sout << "Made ";
      for (auto& op_grad : op_grads) {
        sout << op_grad->Type() << " ";
      }
      VLOG(10) << sout.str();
    }

    for (const auto& desc : op_grads) {
      for (const std::string& out_name : desc->OutputArgumentNames()) {
        if (out_name.find("@GRAD") == std::string::npos) {
          // Not all outputs of a backward operator is a gradient. Only gradient
          // need to be sum. Skip variables are not gradient.
          continue;
        }
        dup_out_ops[out_name].emplace_back(grad_desc_idx);
      }
      ++grad_desc_idx;
    }
    std::transform(op_grads.begin(), op_grads.end(),
                   std::back_inserter(backward_descs),
                   [](std::unique_ptr<OpDesc>& ptr) { return std::move(ptr); });
  }

  VLOG(5) << "Appending Sums";
  // Check whether some variables are written more than once
  std::list<std::pair<size_t, std::unique_ptr<OpDesc>>> pending_sum_ops;
  for (const auto& dup : dup_out_ops) {
    const std::string& out_name = dup.first;
    const std::vector<size_t> dup_op = dup.second;
    if (out_name != kEmptyVarName && dup_op.size() > 1) {
      std::vector<std::string> sum_op_inputs;
      std::string next_g_name = out_name;
      for (size_t i = 0; i < dup_op.size(); ++i) {
        VLOG(10) << backward_descs[dup_op[i]]->Type() << " has " << out_name
                 << " duplicated";
        std::string new_name = out_name + "@RENAME@" + std::to_string(i);
        backward_descs[dup_op[i]]->RenameOutput(out_name, new_name);
        backward_descs[dup_op[i]]->RenameInput(out_name, next_g_name);
        sum_op_inputs.emplace_back(new_name);
        next_g_name = sum_op_inputs.back();
      }
      std::unique_ptr<OpDesc> sum_op(new OpDesc("sum", {{"X", sum_op_inputs}},
                                                {{"Out", {out_name}}},
                                                AttributeMap{}));
      pending_sum_ops.push_back({dup_op.back(), std::move(sum_op)});
    }
  }

  pending_sum_ops.sort([](const std::pair<size_t, std::unique_ptr<OpDesc>>& a,
                          const std::pair<size_t, std::unique_ptr<OpDesc>>& b) {
    return a.first > b.first;
  });
  for (auto& p : pending_sum_ops) {
    backward_descs.insert(backward_descs.begin() + p.first + 1,
                          std::move(p.second));
  }

  VLOG(5) << "MakeBlockBackward Finished";

  return backward_descs;
}

static BlockDesc* CreateStepBlock(
    ProgramDesc& program_desc, std::unordered_set<std::string>* no_grad_vars,
    std::unordered_map<std::string, std::string>* grad_to_var,
    int step_block_idx) {
  auto backward_block_op_descs = MakeBlockBackward(program_desc, step_block_idx,
                                                   no_grad_vars, grad_to_var);
  BlockDesc* backward_block =
      program_desc.AppendBlock(*program_desc.MutableBlock(step_block_idx));
  for (auto& ptr : backward_block_op_descs) {
    backward_block->AppendAllocatedOp(move(ptr));
  }
  return backward_block;
}

ParamGradInfoMap AppendBackward(
    ProgramDesc& program_desc, const VarDesc& target,
    const std::unordered_set<std::string>& no_grad_vars) {
  std::unordered_set<std::string> no_grad_var_names;
  no_grad_var_names.reserve(no_grad_vars.size() + 1);
  no_grad_var_names.insert(std::string(kEmptyVarName) + kGradVarSuffix);
  for (auto& name : no_grad_vars) {
    no_grad_var_names.insert(GradVarName(name));
  }

  const int root_block_idx = 0;
  auto root_block = program_desc.MutableBlock(root_block_idx);

  std::string fill_one_op_out = GradVarName(target.Name());
  bool is_scalar = target.Shape() == std::vector<int64_t>{1};
  PADDLE_ENFORCE(is_scalar, "target should be scalar");
  VLOG(3) << "backward from loss=" << target.Name()
          << " data_type=" << target.GetDataType();
  std::unique_ptr<OpDesc> fill_one_op(
      new OpDesc("fill_constant", {}, {{"Out", {fill_one_op_out}}},
                 {{"shape", std::vector<int>{1}},
                  {"value", static_cast<float>(1.0)},
                  {"dtype", target.GetDataType()}}));
  // infer var type of fill_one_op
  fill_one_op->InferVarType(root_block);

  root_block->AppendAllocatedOp(std::move(fill_one_op));
  size_t forward_op_num = root_block->OpSize();
  size_t forward_block_num = program_desc.Size();

  // Insert backward operators
  std::unordered_map<std::string, std::string> grad_to_var;
  auto backward_op_descs = MakeBlockBackward(program_desc, root_block_idx,
                                             &no_grad_var_names, &grad_to_var);

  for (auto& ptr : backward_op_descs) {
    root_block->AppendAllocatedOp(std::move(ptr));
  }
  // Create Variable

  // Create target gradient variable
  std::unordered_map<std::string, GradVarInfo> retv;

  auto var = root_block->Var(fill_one_op_out);
  var->SetDataType(target.GetDataType());
  var->SetShape(target.Shape());
  auto& target_grad = retv[target.Name()];
  target_grad.name_ = fill_one_op_out;
  target_grad.block_idx_ = root_block_idx;
  target_grad.op_idx_ = static_cast<int>(forward_op_num);

  // create grad_var for all blocks in this program
  CreateGradVarInBlock(forward_op_num, grad_to_var, root_block, &retv);
  for (size_t block_index = forward_block_num;
       block_index < program_desc.Size(); ++block_index) {
    CreateGradVarInBlock(0, grad_to_var, program_desc.MutableBlock(block_index),
                         &retv);
  }
  return retv;
}

}  // namespace framework
}  // namespace paddle