Hendrik/xgboost
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Support slicing tree model (#6302)

Browse Source 
This PR is meant the end the confusion around best_ntree_limit and unify model slicing. We have multi-class and random forests, asking users to understand how to set ntree_limit is difficult and error prone.

* Implement the save_best option in early stopping.

Co-authored-by: Philip Hyunsu Cho <chohyu01@cs.washington.edu>
This commit is contained in:
Jiaming Yuan
2020-11-03 02:27:39 -05:00
committed by GitHub
parent 29745c6df2
commit 2cc9662005
19 changed files with 550 additions and 37 deletions
Download Patch File Download Diff File Expand all files Collapse all files

16
src/c_api/c_api.cc
View File

@@ -730,6 +730,22 @@ XGB_DLL int XGBoosterSaveRabitCheckpoint(BoosterHandle handle) {
API_END();
}
XGB_DLL int XGBoosterSlice(BoosterHandle handle, int begin_layer,
int end_layer, int step,
BoosterHandle *out) {
API_BEGIN();
CHECK_HANDLE();
auto* learner = static_cast<Learner*>(handle);
bool out_of_bound = false;
auto p_out = learner->Slice(begin_layer, end_layer, step, &out_of_bound);
if (out_of_bound) {
return -2;
}
CHECK(p_out);
*out = p_out;
API_END();
}
inline void XGBoostDumpModelImpl(BoosterHandle handle, const FeatureMap &fmap,
int with_stats, const char *format,
xgboost::bst_ulong *len,