Remove update prediction cache from predictors. (#5312)

Move this function into gbtree, and uses only updater for doing so. As now the predictor knows exactly how many trees to predict, there's no need for it to update the prediction cache.

include/xgboost/c_api.h

@@ -531,6 +531,7 @@ XGB_DLL int XGBoosterSaveRabitCheckpoint(BoosterHandle handle);
  *        notice.
  *
  * \param handle handle to Booster object.
+ * \param out_len length of output string
  * \param out_str A valid pointer to array of characters.  The characters array is
  *                allocated and managed by XGBoost, while pointer to that array needs to
  *                be managed by caller.

include/xgboost/gbm.h

@@ -71,7 +71,7 @@ class GradientBooster : public Model, public Configurable {
    * \brief perform update to the model(boosting)
    * \param p_fmat feature matrix that provide access to features
    * \param in_gpair address of the gradient pair statistics of the data
-   * \param obj The objective function, optional, can be nullptr when use customized version
+   * \param prediction The output prediction cache entry that needs to be updated.
    * the booster may change content of gpair
    */
   virtual void DoBoost(DMatrix* p_fmat, HostDeviceVector<GradientPair>* in_gpair,
@@ -158,7 +158,6 @@ class GradientBooster : public Model, public Configurable {
    * \param name name of gradient booster
    * \param generic_param Pointer to runtime parameters
    * \param learner_model_param pointer to global model parameters
-   * \param cache_mats The cache data matrix of the Booster.
    * \return The created booster.
    */
   static GradientBooster* Create(

include/xgboost/generic_parameters.h

@@ -1,6 +1,6 @@
 /*!
  * Copyright 2014-2019 by Contributors
- * \file learner.cc
+ * \file generic_parameters.h
  */
 #ifndef XGBOOST_GENERIC_PARAMETERS_H_
 #define XGBOOST_GENERIC_PARAMETERS_H_

include/xgboost/json.h

@@ -406,7 +406,7 @@ class Json {
   /*! \brief Index Json object with int, used for Json Array. */
   Json& operator[](int ind)                 const { return (*ptr_)[ind]; }
 
-  /*! \Brief Return the reference to stored Json value. */
+  /*! \brief Return the reference to stored Json value. */
   Value const& GetValue() const & { return *ptr_; }
   Value const& GetValue() &&      { return *ptr_; }
   Value&       GetValue() &       { return *ptr_; }

include/xgboost/learner.h

@@ -87,6 +87,7 @@ class Learner : public Model, public Configurable, public rabit::Serializable {
    * \param out_preds output vector that stores the prediction
    * \param ntree_limit limit number of trees used for boosted tree
    *   predictor, when it equals 0, this means we are using all the trees
+   * \param training Whether the prediction result is used for training
    * \param pred_leaf whether to only predict the leaf index of each tree in a boosted tree predictor
    * \param pred_contribs whether to only predict the feature contributions
    * \param approx_contribs whether to approximate the feature contributions for speed

include/xgboost/metric.h

@@ -52,8 +52,9 @@ class Metric {
   /*!
    * \brief create a metric according to name.
    * \param name name of the metric.
-   *  name can be in form metric[@]param
-   *  and the name will be matched in the registry.
+   *        name can be in form metric[@]param and the name will be matched in the
+   *        registry.
+   * \param tparam A global generic parameter
    * \return the created metric.
    */
   static Metric* Create(const std::string& name, GenericParameter const* tparam);

include/xgboost/parameter.h

@@ -1,6 +1,6 @@
 /*!
  * Copyright 2018 by Contributors
- * \file enum_class_param.h
+ * \file parameter.h
  * \brief macro for using C++11 enum class as DMLC parameter
  * \author Hyunsu Philip Cho
  */

include/xgboost/predictor.h

@@ -134,31 +134,6 @@ class Predictor {
                             uint32_t const ntree_limit = 0) = 0;
 
   /**
-   * \fn  virtual void Predictor::UpdatePredictionCache( const gbm::GBTreeModel
-   * &model, std::vector<std::unique_ptr<TreeUpdater> >* updaters, int
-   * num_new_trees) = 0;
-   *
-   * \brief Update the internal prediction cache using newly added trees. Will
-   * use the tree updater to do this if possible. Should be called as a part of
-   * the tree boosting process to facilitate the look up of predictions
-   * at a later time.
-   *
-   * \param           model         The model.
-   * \param [in,out]  updaters      The updater sequence for gradient boosting.
-   * \param           num_new_trees Number of new trees.
-   */
-
-  virtual void UpdatePredictionCache(
-      const gbm::GBTreeModel& model,
-      std::vector<std::unique_ptr<TreeUpdater>>* updaters,
-      int num_new_trees,
-      DMatrix* m,
-      PredictionCacheEntry* predts) = 0;
-
-  /**
-   * \fn  virtual void Predictor::PredictInstance( const SparsePage::Inst&
-   * inst, std::vector<bst_float>* out_preds, const gbm::GBTreeModel& model,
-   *
    * \brief online prediction function, predict score for one instance at a time
    * NOTE: use the batch prediction interface if possible, batch prediction is
    * usually more efficient than online prediction This function is NOT
@@ -234,7 +209,6 @@ class Predictor {
    *
    * \param name           Name of the predictor.
    * \param generic_param  Pointer to runtime parameters.
-   * \param cache          Pointer to prediction cache.
    */
   static Predictor* Create(
       std::string const& name, GenericParameter const* generic_param);

include/xgboost/tree_updater.h

@@ -73,6 +73,7 @@ class TreeUpdater : public Configurable {
   /*!
    * \brief Create a tree updater given name
    * \param name Name of the tree updater.
+   * \param tparam A global runtime parameter
    */
   static TreeUpdater* Create(const std::string& name, GenericParameter const* tparam);
 };