objective_function.hpp

Go to the documentation of this file.

// Copyright (c) 2014-2023, Lawrence Livermore National Security, LLC.
// Produced at the Lawrence Livermore National Laboratory.
// Written by the LBANN Research Team (B. Van Essen, et al.) listed in
// the CONTRIBUTORS file. <lbann-dev@llnl.gov>
//
// LLNL-CODE-697807.
// All rights reserved.
//
// This file is part of LBANN: Livermore Big Artificial Neural Network
// Toolkit. For details, see http://software.llnl.gov/LBANN or
// https://github.com/LLNL/LBANN.
//
// Licensed under the Apache License, Version 2.0 (the "Licensee"); 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.

#ifndef LBANN_OBJECTIVE_FUNCTION_HPP_INCLUDED
#define LBANN_OBJECTIVE_FUNCTION_HPP_INCLUDED

#include "lbann/metrics/metric.hpp"
#include "lbann/objective_functions/objective_function_term.hpp"

namespace lbann {

class objective_function
{
public:
  objective_function() {}

  objective_function(const objective_function& other);
  objective_function& operator=(const objective_function& other);
  ~objective_function() = default;
  objective_function* copy() const { return new objective_function(*this); }

  template <class Archive>
  void serialize(Archive& ar);

  void add_term(std::unique_ptr<objective_function_term> term);
  std::vector<objective_function_term*> get_terms();

  void setup(model& m);

  void start_evaluation(execution_mode mode, int mini_batch_size);

  EvalType finish_evaluation(execution_mode mode, int mini_batch_size);

  void differentiate();

  void compute_weight_regularization();

  void reset_statistics()
  {
    for (auto& stats : m_statistics) {
      stats.second.reset();
    }
  }
  void reset_statistics(execution_mode mode) { m_statistics[mode].reset(); }

  EvalType get_mean_value(execution_mode mode) const;
  int get_statistics_num_samples(execution_mode mode) const;

  std::vector<ViewingLayerPtr> get_layer_pointers() const;
  void set_layer_pointers(std::vector<ViewingLayerPtr> layers);
  std::vector<ViewingWeightsPtr> get_weights_pointers() const;
  void set_weights_pointers(std::vector<ViewingWeightsPtr> w);

  EvalType get_evaluation_time() const { return m_evaluation_time; }
  EvalType get_differentiation_time() const { return m_differentiation_time; }
  void reset_counters()
  {
    m_evaluation_time = 0.0;
    m_differentiation_time = 0.0;
  }

  void write_proto(lbann_data::ObjectiveFunction& proto) const;

private:
  std::vector<std::unique_ptr<objective_function_term>> m_terms;

  std::map<execution_mode, metric_statistics> m_statistics;

  EvalType m_evaluation_time = EvalType(0);
  EvalType m_differentiation_time = EvalType(0);
};

} // namespace lbann

#endif // LBANN_OBJECTIVE_FUNCTION_INCLUDED