Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

[java][mmlspark] Fix cached predictor causing bad values for predicted probabilities #2356

Merged
merged 3 commits into from
Aug 30, 2019
Merged

[java][mmlspark] Fix cached predictor causing bad values for predicted probabilities #2356

Changes from 1 commit
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
f3dd541
[mmlspark] Fix cached predictor causing bad values for predicted prob…
imatiach-msft Aug 25, 2019
7ca8c2e
updated based on comments
imatiach-msft Aug 28, 2019
e0d36a4
removed tabs
imatiach-msft Aug 29, 2019
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Jump to
Jump to file
Failed to load files.
Diff view
Diff view
33 changes: 23 additions & 10 deletions src/c_api.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -46,6 +46,8 @@ catch(std::string& ex) { return LGBM_APIHandleException(ex); } \
catch(...) { return LGBM_APIHandleException("unknown exception"); } \
return 0;

const int PREDICTOR_TYPES = 4;

class Booster {
public:
explicit Booster(const char* filename) {
Expand Down Expand Up @@ -205,8 +207,15 @@ class Booster {
const Config& config,
double* out_result, int64_t* out_len) {
std::lock_guard<std::mutex> lock(mutex_);

if (single_row_predictor_.get() == nullptr) {
if (single_row_predictor_[predict_type].get() == nullptr ||
single_row_predictor_pred_early_stop[predict_type] != config.pred_early_stop ||
single_row_predictor_pred_early_stop_freq[predict_type] != config.pred_early_stop_freq ||
single_row_predictor_pred_early_stop_margin[predict_type] != config.pred_early_stop_margin ||
single_row_predictor_iter_[predict_type] != num_iteration) {
single_row_predictor_iter_[predict_type] = num_iteration;
single_row_predictor_pred_early_stop[predict_type] = config.pred_early_stop;
single_row_predictor_pred_early_stop_freq[predict_type] = config.pred_early_stop_freq;
single_row_predictor_pred_early_stop_margin[predict_type] = config.pred_early_stop_margin;
imatiach-msft marked this conversation as resolved.
Show resolved Hide resolved
bool is_predict_leaf = false;
bool is_raw_score = false;
bool predict_contrib = false;
Expand All @@ -221,17 +230,17 @@ class Booster {
}

// TODO(eisber): config could be optimized away... (maybe using lambda callback?)
single_row_predictor_.reset(new Predictor(boosting_.get(), num_iteration, is_raw_score, is_predict_leaf, predict_contrib,
single_row_predictor_[predict_type].reset(new Predictor(boosting_.get(), num_iteration, is_raw_score, is_predict_leaf, predict_contrib,
config.pred_early_stop, config.pred_early_stop_freq, config.pred_early_stop_margin));
single_row_num_pred_in_one_row_ = boosting_->NumPredictOneRow(num_iteration, is_predict_leaf, predict_contrib);
single_row_predict_function_ = single_row_predictor_->GetPredictFunction();
single_row_num_pred_in_one_row_[predict_type] = boosting_->NumPredictOneRow(num_iteration, is_predict_leaf, predict_contrib);
single_row_predict_function_[predict_type] = single_row_predictor_[predict_type]->GetPredictFunction();
}

auto one_row = get_row_fun(0);
auto pred_wrt_ptr = out_result;
single_row_predict_function_(one_row, pred_wrt_ptr);
single_row_predict_function_[predict_type](one_row, pred_wrt_ptr);

*out_len = single_row_num_pred_in_one_row_;
*out_len = single_row_num_pred_in_one_row_[predict_type];
}


Expand Down Expand Up @@ -364,9 +373,13 @@ class Booster {
private:
const Dataset* train_data_;
std::unique_ptr<Boosting> boosting_;
std::unique_ptr<Predictor> single_row_predictor_;
PredictFunction single_row_predict_function_;
int64_t single_row_num_pred_in_one_row_;
std::unique_ptr<Predictor> single_row_predictor_[PREDICTOR_TYPES];
bool single_row_predictor_pred_early_stop[PREDICTOR_TYPES];
int single_row_predictor_pred_early_stop_freq[PREDICTOR_TYPES];
Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

just to confirm, will these arrays be freed correctly with the class on the heap once class is deleted?

Copy link
Contributor

@eisber eisber Aug 28, 2019 •
edited

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

those are fixed size arrays and they'll be deallocated directly.

I'd suggest to use create a

  • new class containing the relevant subset (unique_ptr, bool, int)
  • implement operator== (const Config&) and operator=(const Config&)
  • would make the assignment and comparison a bit more readable.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@eisber I already created a new class containing those variables, can you take a look at the latest commit?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

also for comparison "implement operator== (const Config&) and operator=(const Config&)" I created a method since I needed to compare other variables besides the Config

double single_row_predictor_pred_early_stop_margin[PREDICTOR_TYPES];
int single_row_predictor_iter_[PREDICTOR_TYPES];
imatiach-msft marked this conversation as resolved.
Show resolved Hide resolved
PredictFunction single_row_predict_function_[PREDICTOR_TYPES];
int64_t single_row_num_pred_in_one_row_[PREDICTOR_TYPES];
imatiach-msft marked this conversation as resolved.
Show resolved Hide resolved

/*! \brief All configs */
Config config_;
Expand Down