[python] [R-package] Use the same address when updated label/weight/query (#2662)

* Update metadata.cpp

* add version for training set, for efficiently update label/weight/... during training.

* Update lgb.Booster.R

R-package/R/lgb.Booster.R

@@ -55,6 +55,7 @@ Booster <- R6::R6Class(
 
           # Create private booster information
           private$train_set <- train_set
+          private$train_set_version <- train_set$.__enclos_env__$private$version
           private$num_dataset <- 1L
           private$init_predictor <- train_set$.__enclos_env__$private$predictor
 
@@ -207,6 +208,12 @@ Booster <- R6::R6Class(
     # Perform boosting update iteration
     update = function(train_set = NULL, fobj = NULL) {
 
+      if (is.null(train_set)) {
+        if (private$train_set$.__enclos_env__$private$version != private$train_set_version) {
+          train_set <- private$train_set
+        }
+      }
+
       # Check if training set is not null
       if (!is.null(train_set)) {
 
@@ -230,6 +237,7 @@ Booster <- R6::R6Class(
 
         # Store private train set
         private$train_set <- train_set
+        private$train_set_version <- train_set$.__enclos_env__$private$version
 
       }
 
@@ -497,6 +505,7 @@ Booster <- R6::R6Class(
     eval_names = NULL,
     higher_better_inner_eval = NULL,
     set_objective_to_none = FALSE,
+    train_set_version = 0L,
     # Predict data
     inner_predict = function(idx) {
 

R-package/R/lgb.Dataset.R

@@ -89,6 +89,7 @@ Dataset <- R6::R6Class(
       private$free_raw_data <- free_raw_data
       private$used_indices <- sort(used_indices, decreasing = FALSE)
       private$info <- info
+      private$version <- 0L
 
     },
 
@@ -503,6 +504,8 @@ Dataset <- R6::R6Class(
             , length(info)
           )
 
+          private$version <- private$version + 1L
+
         }
 
       }
@@ -638,6 +641,7 @@ Dataset <- R6::R6Class(
     free_raw_data = TRUE,
     used_indices = NULL,
     info = NULL,
+    version = 0L,
 
     # Get handle
     get_handle = function() {

python-package/lightgbm/basic.py

@@ -771,6 +771,7 @@ class Dataset(object):
         self.params_back_up = None
         self.feature_penalty = None
         self.monotone_constraints = None
+        self.version = 0
 
     def __del__(self):
         try:
@@ -1233,6 +1234,7 @@ class Dataset(object):
             ptr_data,
             ctypes.c_int(len(data)),
             ctypes.c_int(type_data)))
+        self.version += 1
         return self
 
     def get_field(self, field_name):
@@ -1740,6 +1742,7 @@ class Booster(object):
             self.__is_predicted_cur_iter = [False]
             self.__get_eval_info()
             self.pandas_categorical = train_set.pandas_categorical
+            self.train_set_version = train_set.version
         elif model_file is not None:
             # Prediction task
             out_num_iterations = ctypes.c_int(0)
@@ -2076,7 +2079,12 @@ class Booster(object):
             Whether the update was successfully finished.
         """
         # need reset training data
-        if train_set is not None and train_set is not self.train_set:
+        if train_set is None and self.train_set_version != self.train_set.version:
+            train_set = self.train_set
+            is_the_same_train_set = False
+        else:
+            is_the_same_train_set = train_set is self.train_set and self.train_set_version == train_set.version
+        if train_set is not None and not is_the_same_train_set:
             if not isinstance(train_set, Dataset):
                 raise TypeError('Training data should be Dataset instance, met {}'
                                 .format(type(train_set).__name__))
@@ -2088,6 +2096,7 @@ class Booster(object):
                 self.handle,
                 self.train_set.construct().handle))
             self.__inner_predict_buffer[0] = None
+            self.train_set_version = self.train_set.version
         is_finished = ctypes.c_int(0)
         if fobj is None:
             if self.__set_objective_to_none:

src/io/metadata.cpp

@@ -290,9 +290,9 @@ void Metadata::SetInitScore(const double* init_score, data_size_t len) {
   if ((len % num_data_) != 0) {
     Log::Fatal("Initial score size doesn't match data size");
   }
-  if (!init_score_.empty()) { init_score_.clear(); }
+  if (init_score_.empty()) { init_score_.resize(len); }
   num_init_score_ = len;
-  init_score_ = std::vector<double>(len);
+
   #pragma omp parallel for schedule(static)
   for (int64_t i = 0; i < num_init_score_; ++i) {
     init_score_[i] = Common::AvoidInf(init_score[i]);
@@ -308,8 +308,8 @@ void Metadata::SetLabel(const label_t* label, data_size_t len) {
   if (num_data_ != len) {
     Log::Fatal("Length of label is not same with #data");
   }
-  if (!label_.empty()) { label_.clear(); }
-  label_ = std::vector<label_t>(num_data_);
+  if (label_.empty()) { label_.resize(num_data_); }
+ 
   #pragma omp parallel for schedule(static)
   for (data_size_t i = 0; i < num_data_; ++i) {
     label_[i] = Common::AvoidInf(label[i]);
@@ -327,9 +327,9 @@ void Metadata::SetWeights(const label_t* weights, data_size_t len) {
   if (num_data_ != len) {
     Log::Fatal("Length of weights is not same with #data");
   }
-  if (!weights_.empty()) { weights_.clear(); }
+  if (weights_.empty()) { weights_.resize(num_data_); }
   num_weights_ = num_data_;
-  weights_ = std::vector<label_t>(num_weights_);
+ 
   #pragma omp parallel for schedule(static)
   for (data_size_t i = 0; i < num_weights_; ++i) {
     weights_[i] = Common::AvoidInf(weights[i]);
@@ -354,9 +354,8 @@ void Metadata::SetQuery(const data_size_t* query, data_size_t len) {
   if (num_data_ != sum) {
     Log::Fatal("Sum of query counts is not same with #data");
   }
-  if (!query_boundaries_.empty()) { query_boundaries_.clear(); }
   num_queries_ = len;
-  query_boundaries_ = std::vector<data_size_t>(num_queries_ + 1);
+  query_boundaries_.resize(num_queries_ + 1);
   query_boundaries_[0] = 0;
   for (data_size_t i = 0; i < num_queries_; ++i) {
     query_boundaries_[i + 1] = query_boundaries_[i] + query[i];