Remove experimental_json_serialization from tests. (#6640)

src/learner.cc

@@ -632,8 +632,7 @@ class LearnerIO : public LearnerConfiguration {
  private:
   std::set<std::string> saved_configs_ = {"num_round"};
   // Used to identify the offset of JSON string when
-  // `enable_experimental_json_serialization' is set to false.  Will be removed once JSON
+  // Will be removed once JSON takes over.  Right now we still loads some RDS files from R.
-  // takes over.
   std::string const serialisation_header_ { u8"CONFIG-offset:" };
 
  public:
@@ -816,10 +815,7 @@ class LearnerIO : public LearnerConfiguration {
   }
 
   // Save model into binary format.  The code is about to be deprecated by more robust
-  // JSON serialization format.  This function is uneffected by
+  // JSON serialization format.
-  // `enable_experimental_json_serialization` as user might enable this flag for pickle
-  // while still want a binary output.  As we are progressing at replacing the binary
-  // format, there's no need to put too much effort on it.
   void SaveModel(dmlc::Stream* fo) const override {
     LearnerModelParamLegacy mparam = mparam_;  // make a copy to potentially modify
     std::vector<std::pair<std::string, std::string> > extra_attr;

tests/cpp/test_serialization.cc

@@ -339,7 +339,6 @@ TEST_F(SerializationTest, ConfigurationCount) {
     auto learner = std::unique_ptr<Learner>(Learner::Create(mat));
 
     learner->SetParam("tree_method", "gpu_hist");
-    learner->SetParam("enable_experimental_json_serialization", "1");
 
     for (size_t i = 0; i < 10; ++i) {
       learner->UpdateOneIter(i, p_dmat);

tests/python-gpu/test_gpu_prediction.py

@@ -283,9 +283,10 @@ class TestGPUPredict:
         y = (x0 * 10 - 20) + (x1 - 2)
         dtrain = xgb.DMatrix(df, label=y, enable_categorical=True)
 
-        params = {'tree_method': 'gpu_hist', 'predictor': 'gpu_predictor',
+        params = {
-                  'enable_experimental_json_serialization': True,
+            'tree_method': 'gpu_hist', 'predictor': 'gpu_predictor',
-                  'max_depth': 3, 'learning_rate': 1.0, 'base_score': 0.0, 'eval_metric': 'rmse'}
+            'max_depth': 3, 'learning_rate': 1.0, 'base_score': 0.0, 'eval_metric': 'rmse'
+        }
 
         eval_history = {}
         bst = xgb.train(params, dtrain, num_boost_round=5, evals=[(dtrain, 'train')],

tests/python-gpu/test_gpu_training_continuation.py

@@ -6,15 +6,14 @@ rng = np.random.RandomState(1994)
 
 
 class TestGPUTrainingContinuation:
-    def run_training_continuation(self, use_json):
+    def test_training_continuation(self):
         kRows = 64
         kCols = 32
         X = np.random.randn(kRows, kCols)
         y = np.random.randn(kRows)
         dtrain = xgb.DMatrix(X, y)
         params = {'tree_method': 'gpu_hist', 'max_depth': '2',
-                  'gamma': '0.1', 'alpha': '0.01',
+                  'gamma': '0.1', 'alpha': '0.01'}
-                  'enable_experimental_json_serialization': use_json}
         bst_0 = xgb.train(params, dtrain, num_boost_round=64)
         dump_0 = bst_0.get_dump(dump_format='json')
 
@@ -48,9 +47,3 @@ class TestGPUTrainingContinuation:
             obj_0 = json.loads(dump_0[i])
             obj_1 = json.loads(dump_1[i])
             recursive_compare(obj_0, obj_1)
-
-    def test_gpu_training_continuation_binary(self):
-        self.run_training_continuation(False)
-
-    def test_gpu_training_continuation_json(self):
-        self.run_training_continuation(True)

tests/python-gpu/test_gpu_updaters.py

@@ -63,9 +63,7 @@ class TestGPUUpdaters:
         by_etl_results = {}
         by_builtin_results = {}
 
-        parameters = {'tree_method': 'gpu_hist',
+        parameters = {'tree_method': 'gpu_hist', 'predictor': 'gpu_predictor'}
-                      'predictor': 'gpu_predictor',
-                      'enable_experimental_json_serialization': True}
 
         m = xgb.DMatrix(onehot, label, enable_categorical=True)
         xgb.train(parameters, m,