fix predict batch size in scikit

Xtra-Computing · Jul 14, 2018 · cae0a26 · cae0a26
1 parent 07c812e
commit cae0a26
Showing 1 changed file with 2 additions and 2 deletions.
diff --git a/src/thundersvm/thundersvm-scikit.cpp b/src/thundersvm/thundersvm-scikit.cpp
@@ -289,7 +289,7 @@ extern "C" {
     void sparse_decision(int row_size, float* val, int* row_ptr, int* col_ptr, SvmModel *model, int value_size, float* dec_value){
         DataSet predict_dataset;
         predict_dataset.load_from_sparse(row_size, val, row_ptr, col_ptr, (float *)NULL);
-        model->predict(predict_dataset.instances(), 10000);
+        model->predict(predict_dataset.instances(), -1);
         SyncArray<float_type> dec_value_array(value_size);
         dec_value_array.copy_from(model->get_dec_value());
         float_type *dec_value_ptr = dec_value_array.host_data();
@@ -301,7 +301,7 @@ extern "C" {
     void dense_decision(int row_size, int features, float* data, SvmModel *model, int value_size, float* dec_value){
         DataSet predict_dataset;
         predict_dataset.load_from_dense(row_size, features, data, (float*) NULL);
-        model->predict(predict_dataset.instances(), 10000);
+        model->predict(predict_dataset.instances(), -1);
         SyncArray<float_type> dec_value_array(value_size);
         dec_value_array.copy_from(model->get_dec_value());
         float_type *dec_value_ptr = dec_value_array.host_data();