nnet-discriminative-diagnostics.cc

Go to the documentation of this file.

// nnet3/nnet-discriminative-diagnostics.cc

// Copyright  2012-2015    Johns Hopkins University (author: Daniel Povey)
// Copyright  2014-2015    Vimal Manohar

// See ../../COPYING for clarification regarding multiple authors
//
// 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
//
// THIS CODE IS PROVIDED *AS IS* BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
// KIND, EITHER EXPRESS OR IMPLIED, INCLUDING WITHOUT LIMITATION ANY IMPLIED
// WARRANTIES OR CONDITIONS OF TITLE, FITNESS FOR A PARTICULAR PURPOSE,
// MERCHANTABLITY OR NON-INFRINGEMENT.
// See the Apache 2 License for the specific language governing permissions and
// limitations under the License.

#include "nnet3/nnet-discriminative-diagnostics.h"
#include "nnet3/nnet-utils.h"
#include "nnet3/discriminative-training.h"

namespace kaldi {
namespace nnet3 {

NnetDiscriminativeComputeObjf::NnetDiscriminativeComputeObjf(
    const NnetComputeProbOptions &nnet_config,
    const discriminative::DiscriminativeOptions &discriminative_config,
    const TransitionModel &tmodel,
    const VectorBase<BaseFloat> &priors,
    const Nnet &nnet):
    nnet_config_(nnet_config),
    discriminative_config_(discriminative_config),
    tmodel_(tmodel),
    log_priors_(priors),
    nnet_(nnet),
    compiler_(nnet, nnet_config_.optimize_config),
    deriv_nnet_(NULL),
    num_minibatches_processed_(0) {
  log_priors_.ApplyLog();
  if (nnet_config_.compute_deriv) {
    deriv_nnet_ = new Nnet(nnet_);
    ScaleNnet(0.0, deriv_nnet_);
    SetNnetAsGradient(deriv_nnet_); // force simple update
  }
}

const Nnet& NnetDiscriminativeComputeObjf::GetDeriv() const {
  if (deriv_nnet_ == NULL)
    KALDI_ERR << "GetDeriv() called when no derivatives were requested.";
  return *deriv_nnet_;
}

NnetDiscriminativeComputeObjf::~NnetDiscriminativeComputeObjf() {
  delete deriv_nnet_;  // delete does nothing if pointer is NULL.
}

void NnetDiscriminativeComputeObjf::Reset() {
  num_minibatches_processed_ = 0;
  objf_info_.clear();
  if (deriv_nnet_) {
    ScaleNnet(0.0, deriv_nnet_);
    SetNnetAsGradient(deriv_nnet_);
  }
}

void NnetDiscriminativeComputeObjf::Compute(const NnetDiscriminativeExample &eg) {
  bool need_model_derivative = nnet_config_.compute_deriv,
      store_component_stats = false;
  bool use_xent_regularization = (discriminative_config_.xent_regularize != 0.0),
      use_xent_derivative = false;

  ComputationRequest request;
  GetDiscriminativeComputationRequest(nnet_, eg,
                                      need_model_derivative,
                                      store_component_stats,
                                      use_xent_regularization, use_xent_derivative,
                                      &request);
  std::shared_ptr<const NnetComputation> computation = compiler_.Compile(request);
  NnetComputer computer(nnet_config_.compute_config, *computation,
                        nnet_, deriv_nnet_);
  // give the inputs to the computer object.
  computer.AcceptInputs(nnet_, eg.inputs);
  computer.Run();
  this->ProcessOutputs(eg, &computer);
  if (nnet_config_.compute_deriv)
    computer.Run();
}

void NnetDiscriminativeComputeObjf::ProcessOutputs(
                                    const NnetDiscriminativeExample &eg,
                                    NnetComputer *computer) {
  // There will normally be just one output here, named 'output',
  // but the code is more general than this.
  std::vector<NnetDiscriminativeSupervision>::const_iterator iter = eg.outputs.begin(),
      end = eg.outputs.end();
  for (; iter != end; ++iter) {
    const NnetDiscriminativeSupervision &sup = *iter;
    int32 node_index = nnet_.GetNodeIndex(sup.name);
    if (node_index < 0 ||
        !nnet_.IsOutputNode(node_index))
      KALDI_ERR << "Network has no output named " << sup.name;

    const CuMatrixBase<BaseFloat> &nnet_output = computer->GetOutput(sup.name);

    bool use_xent = (discriminative_config_.xent_regularize != 0.0);
    std::string xent_name = sup.name + "-xent";  // typically "output-xent".
    CuMatrix<BaseFloat> nnet_output_deriv, xent_deriv;

    if (nnet_config_.compute_deriv)
      nnet_output_deriv.Resize(nnet_output.NumRows(), nnet_output.NumCols(),
                               kUndefined);

    if (use_xent)
      xent_deriv.Resize(nnet_output.NumRows(), nnet_output.NumCols(),
                        kUndefined);

    if (objf_info_.count(sup.name) == 0)
      objf_info_.insert(std::make_pair(sup.name,
          discriminative::DiscriminativeObjectiveInfo(discriminative_config_)));

    discriminative::DiscriminativeObjectiveInfo *stats = &(objf_info_[sup.name]);

    discriminative::ComputeDiscriminativeObjfAndDeriv(discriminative_config_,
                                                      tmodel_, log_priors_,
                                                      sup.supervision, nnet_output,
                                                      stats,
                                                      (nnet_config_.compute_deriv ?
                                                       &nnet_output_deriv : NULL),
                                                      (use_xent ? &xent_deriv : NULL));

    if (nnet_config_.compute_deriv)
      computer->AcceptInput(sup.name, &nnet_output_deriv);

    if (use_xent) {
      if (objf_info_.count(xent_name) == 0)
        objf_info_.insert(std::make_pair(xent_name,
          discriminative::DiscriminativeObjectiveInfo(discriminative_config_)));
      discriminative::DiscriminativeObjectiveInfo &xent_stats = objf_info_[xent_name];

      // this block computes the cross-entropy objective.
      const CuMatrixBase<BaseFloat> &xent_output = computer->GetOutput(xent_name);
      // at this point, xent_deriv is posteriors derived from the numerator
      // computation.  note, xent_deriv has a factor of 'supervision.weight',
      // but so does tot_weight.
      BaseFloat xent_objf = TraceMatMat(xent_output, xent_deriv, kTrans);
      xent_stats.tot_t_weighted += stats->tot_t_weighted;
      xent_stats.tot_objf += xent_objf;
    }

    num_minibatches_processed_++;
  }
}

bool NnetDiscriminativeComputeObjf::PrintTotalStats() const {
  bool ans = false;
  unordered_map<std::string, discriminative::DiscriminativeObjectiveInfo, StringHasher>::const_iterator
      iter, end;
  iter = objf_info_.begin();
  end = objf_info_.end();
  for (; iter != end; ++iter) {
    const std::string &name = iter->first;
    int32 node_index = nnet_.GetNodeIndex(name);
    KALDI_ASSERT(node_index >= 0);
    const discriminative::DiscriminativeObjectiveInfo &info = iter->second;
    BaseFloat tot_weight = info.tot_t_weighted;
    BaseFloat tot_objective = info.TotalObjf(
        discriminative_config_.criterion);

    info.PrintAll(discriminative_config_.criterion);

    if (info.tot_l2_term == 0.0) {
      KALDI_LOG << "Overall " << discriminative_config_.criterion
                << " objective for '"
                << name << "' is "
                << (tot_objective / tot_weight)
                << " per frame, "
                << "over " << tot_weight << " frames.";
    } else {
      KALDI_LOG << "Overall " << discriminative_config_.criterion
                << " objective for '"
                << name << "' is "
                << (tot_objective / tot_weight)
                << " + " << (info.tot_l2_term / tot_weight)
                << " per frame, "
                << "over " << tot_weight << " frames.";
    }

    if (tot_weight > 0)
      ans = true;
  }
  return ans;
}

const discriminative::DiscriminativeObjectiveInfo* NnetDiscriminativeComputeObjf::GetObjective(
    const std::string &output_name) const {
  unordered_map<std::string, discriminative::DiscriminativeObjectiveInfo, StringHasher>::const_iterator
      iter = objf_info_.find(output_name);
  if (iter != objf_info_.end())
    return &(iter->second);
  else
    return NULL;
}

} // namespace nnet3
} // namespace kaldi