Comments after code review

src/boosting/gbdt.cpp

@@ -290,8 +290,10 @@ void GBDT::RefitTree(const int* tree_leaf_prediction, const size_t nrow, const s
 }
 
 void GBDT::RefitTreeManual(int tree_idx, const double *vals, const int vals_size) {
-  CHECK(tree_idx >= 0 && static_cast<size_t>(tree_idx) < models_.size() && vals_size == models_[tree_idx]->num_leaves());
-  // reset score
+  CHECK(tree_idx >= 0);
+  CHECK(static_cast<size_t>(tree_idx) < models_.size());
+  CHECK(vals_size == models_[tree_idx]->num_leaves());
+  // reset score by adding the difference
   for (int leaf_id = 0; leaf_id < models_[tree_idx]->num_leaves(); ++leaf_id) {
     models_[tree_idx]->SetLeafOutput(leaf_id, vals[leaf_id] - models_[tree_idx]->LeafOutput(leaf_id));
   }

tests/python_package_test/test_engine.py

@@ -2366,9 +2366,13 @@ def debias_callback(env):
         "num_leaves": 5,
         "objective": "gamma",
     }
-    bst = lgb.train(params, ds, callbacks=[debias_callback])
+
+    # Check that the model is biased when no callback is provided
+    bst = lgb.train(params, ds)
+    np.testing.assert_raises(AssertionError, np.testing.assert_array_equal, bst.predict(df).mean(), y.mean())
 
     # Check if debiasing worked
+    bst = lgb.train(params, ds, callbacks=[debias_callback])
     np.testing.assert_allclose(bst.predict(df).mean(), y.mean())