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

[Breaking] Accept multi-dim meta info. (#7405)

Browse Source 
This PR changes base_margin into a 3-dim array, with one of them being reserved for multi-target classification. Also, a breaking change is made for binary serialization due to extra dimension along with a fix for saving the feature weights. Lastly, it unifies the prediction initialization between CPU and GPU. After this PR, the meta info setter in Python will be based on array interface.
This commit is contained in:
Jiaming Yuan
2021-11-18 23:02:54 +08:00
committed by GitHub
parent 9fb4338964
commit d33854af1b
25 changed files with 545 additions and 256 deletions
Download Patch File Download Diff File Expand all files Collapse all files

28
src/predictor/cpu_predictor.cc
View File

@@ -282,27 +282,6 @@ class CPUPredictor : public Predictor {
}
}
void InitOutPredictions(const MetaInfo& info,
HostDeviceVector<bst_float>* out_preds,
const gbm::GBTreeModel& model) const override {
CHECK_NE(model.learner_model_param->num_output_group, 0);
size_t n = model.learner_model_param->num_output_group * info.num_row_;
const auto& base_margin = info.base_margin_.HostVector();
out_preds->Resize(n);
std::vector<bst_float>& out_preds_h = out_preds->HostVector();
if (base_margin.empty()) {
std::fill(out_preds_h.begin(), out_preds_h.end(),
model.learner_model_param->base_score);
} else {
std::string expected{
"(" + std::to_string(info.num_row_) + ", " +
std::to_string(model.learner_model_param->num_output_group) + ")"};
CHECK_EQ(base_margin.size(), n)
<< "Invalid shape of base_margin. Expected:" << expected;
std::copy(base_margin.begin(), base_margin.end(), out_preds_h.begin());
}
}
public:
explicit CPUPredictor(GenericParameter const* generic_param) :
Predictor::Predictor{generic_param} {}
@@ -456,7 +435,7 @@ class CPUPredictor : public Predictor {
common::ParallelFor(bst_omp_uint(ntree_limit), [&](bst_omp_uint i) {
FillNodeMeanValues(model.trees[i].get(), &(mean_values[i]));
});
const std::vector<bst_float>& base_margin = info.base_margin_.HostVector();
auto base_margin = info.base_margin_.View(GenericParameter::kCpuId);
// start collecting the contributions
for (const auto &batch : p_fmat->GetBatches<SparsePage>()) {
auto page = batch.GetView();
@@ -496,8 +475,9 @@ class CPUPredictor : public Predictor {
}
feats.Drop(page[i]);
// add base margin to BIAS
if (base_margin.size() != 0) {
p_contribs[ncolumns - 1] += base_margin[row_idx * ngroup + gid];
if (base_margin.Size() != 0) {
CHECK_EQ(base_margin.Shape(1), ngroup);
p_contribs[ncolumns - 1] += base_margin(row_idx, gid);
} else {
p_contribs[ncolumns - 1] += model.learner_model_param->base_score;
}

25
src/predictor/gpu_predictor.cu
View File

@@ -855,7 +855,7 @@ class GPUPredictor : public xgboost::Predictor {
}
// Add the base margin term to last column
p_fmat->Info().base_margin_.SetDevice(generic_param_->gpu_id);
const auto margin = p_fmat->Info().base_margin_.ConstDeviceSpan();
const auto margin = p_fmat->Info().base_margin_.Data()->ConstDeviceSpan();
float base_score = model.learner_model_param->base_score;
dh::LaunchN(
p_fmat->Info().num_row_ * model.learner_model_param->num_output_group,
@@ -914,7 +914,7 @@ class GPUPredictor : public xgboost::Predictor {
}
// Add the base margin term to last column
p_fmat->Info().base_margin_.SetDevice(generic_param_->gpu_id);
const auto margin = p_fmat->Info().base_margin_.ConstDeviceSpan();
const auto margin = p_fmat->Info().base_margin_.Data()->ConstDeviceSpan();
float base_score = model.learner_model_param->base_score;
size_t n_features = model.learner_model_param->num_feature;
dh::LaunchN(
@@ -928,27 +928,6 @@ class GPUPredictor : public xgboost::Predictor {
});
}
protected:
void InitOutPredictions(const MetaInfo& info,
HostDeviceVector<bst_float>* out_preds,
const gbm::GBTreeModel& model) const override {
size_t n_classes = model.learner_model_param->num_output_group;
size_t n = n_classes * info.num_row_;
const HostDeviceVector<bst_float>& base_margin = info.base_margin_;
out_preds->SetDevice(generic_param_->gpu_id);
out_preds->Resize(n);
if (base_margin.Size() != 0) {
std::string expected{
"(" + std::to_string(info.num_row_) + ", " +
std::to_string(model.learner_model_param->num_output_group) + ")"};
CHECK_EQ(base_margin.Size(), n)
<< "Invalid shape of base_margin. Expected:" << expected;
out_preds->Copy(base_margin);
} else {
out_preds->Fill(model.learner_model_param->base_score);
}
}
void PredictInstance(const SparsePage::Inst&,
std::vector<bst_float>*,
const gbm::GBTreeModel&, unsigned) const override {

36
src/predictor/predictor.cc
View File

@@ -1,5 +1,5 @@
/*!
* Copyright 2017-2020 by Contributors
* Copyright 2017-2021 by Contributors
*/
#include <dmlc/registry.h>
#include <mutex>
@@ -8,6 +8,8 @@
#include "xgboost/data.h"
#include "xgboost/generic_parameters.h"
#include "../gbm/gbtree.h"
namespace dmlc {
DMLC_REGISTRY_ENABLE(::xgboost::PredictorReg);
} // namespace dmlc
@@ -58,6 +60,38 @@ Predictor* Predictor::Create(
auto p_predictor = (e->body)(generic_param);
return p_predictor;
}
void ValidateBaseMarginShape(linalg::Tensor<float, 3> const& margin, bst_row_t n_samples,
bst_group_t n_groups) {
// FIXME: Bindings other than Python doesn't have shape.
std::string expected{"Invalid shape of base_margin. Expected: (" + std::to_string(n_samples) +
", " + std::to_string(n_groups) + ")"};
CHECK_EQ(margin.Shape(0), n_samples) << expected;
CHECK_EQ(margin.Shape(1), n_groups) << expected;
}
void Predictor::InitOutPredictions(const MetaInfo& info, HostDeviceVector<bst_float>* out_preds,
const gbm::GBTreeModel& model) const {
CHECK_NE(model.learner_model_param->num_output_group, 0);
size_t n_classes = model.learner_model_param->num_output_group;
size_t n = n_classes * info.num_row_;
const HostDeviceVector<bst_float>* base_margin = info.base_margin_.Data();
if (generic_param_->gpu_id >= 0) {
out_preds->SetDevice(generic_param_->gpu_id);
}
if (base_margin->Size() != 0) {
out_preds->Resize(n);
ValidateBaseMarginShape(info.base_margin_, info.num_row_, n_classes);
out_preds->Copy(*base_margin);
} else {
if (out_preds->Empty()) {
out_preds->Resize(n, model.learner_model_param->base_score);
} else {
out_preds->Resize(n);
out_preds->Fill(model.learner_model_param->base_score);
}
}
}
} // namespace xgboost
namespace xgboost {