Fix mixed types in GPU sketching. (#7228)
This commit is contained in:
parent
037dd0820d
commit
2942dc68e4
@ -156,22 +156,24 @@ void RemoveDuplicatedCategories(
|
|||||||
auto d_old_column_sizes_scan = dh::ToSpan(column_sizes_scan);
|
auto d_old_column_sizes_scan = dh::ToSpan(column_sizes_scan);
|
||||||
dh::caching_device_vector<SketchContainer::OffsetT> new_cuts_size(
|
dh::caching_device_vector<SketchContainer::OffsetT> new_cuts_size(
|
||||||
info.num_col_ + 1);
|
info.num_col_ + 1);
|
||||||
auto d_new_cuts_size = dh::ToSpan(new_cuts_size);
|
|
||||||
auto d_new_columns_ptr = dh::ToSpan(new_column_scan);
|
|
||||||
CHECK_EQ(new_column_scan.size(), new_cuts_size.size());
|
CHECK_EQ(new_column_scan.size(), new_cuts_size.size());
|
||||||
dh::LaunchN(new_column_scan.size(), [=] __device__(size_t idx) {
|
dh::LaunchN(
|
||||||
d_old_column_sizes_scan[idx] = d_new_columns_ptr[idx];
|
new_column_scan.size(),
|
||||||
if (idx == d_new_columns_ptr.size() - 1) {
|
[=, d_new_cuts_size = dh::ToSpan(new_cuts_size),
|
||||||
return;
|
d_old_column_sizes_scan = dh::ToSpan(column_sizes_scan),
|
||||||
}
|
d_new_columns_ptr = dh::ToSpan(new_column_scan)] __device__(size_t idx) {
|
||||||
if (IsCat(d_feature_types, idx)) {
|
d_old_column_sizes_scan[idx] = d_new_columns_ptr[idx];
|
||||||
// Cut size is the same as number of categories in input.
|
if (idx == d_new_columns_ptr.size() - 1) {
|
||||||
d_new_cuts_size[idx] =
|
return;
|
||||||
d_new_columns_ptr[idx + 1] - d_new_columns_ptr[idx];
|
}
|
||||||
} else {
|
if (IsCat(d_feature_types, idx)) {
|
||||||
d_new_cuts_size[idx] = d_cuts_ptr[idx] - d_cuts_ptr[idx];
|
// Cut size is the same as number of categories in input.
|
||||||
}
|
d_new_cuts_size[idx] =
|
||||||
});
|
d_new_columns_ptr[idx + 1] - d_new_columns_ptr[idx];
|
||||||
|
} else {
|
||||||
|
d_new_cuts_size[idx] = d_cuts_ptr[idx + 1] - d_cuts_ptr[idx];
|
||||||
|
}
|
||||||
|
});
|
||||||
// Turn size into ptr.
|
// Turn size into ptr.
|
||||||
thrust::exclusive_scan(thrust::device, new_cuts_size.cbegin(),
|
thrust::exclusive_scan(thrust::device, new_cuts_size.cbegin(),
|
||||||
new_cuts_size.cend(), d_cuts_ptr.data());
|
new_cuts_size.cend(), d_cuts_ptr.data());
|
||||||
|
|||||||
@ -165,6 +165,32 @@ TEST(HistUtil, DeviceSketchCategoricalFeatures) {
|
|||||||
TestCategoricalSketch(1000, 256, 32, true);
|
TestCategoricalSketch(1000, 256, 32, true);
|
||||||
}
|
}
|
||||||
|
|
||||||
|
void TestMixedSketch() {
|
||||||
|
size_t n_samples = 1000, n_features = 2, n_categories = 3;
|
||||||
|
std::vector<float> data(n_samples * n_features);
|
||||||
|
SimpleLCG gen;
|
||||||
|
SimpleRealUniformDistribution<float> cat_d{0.0f, float(n_categories)};
|
||||||
|
SimpleRealUniformDistribution<float> num_d{0.0f, 3.0f};
|
||||||
|
for (size_t i = 0; i < n_samples * n_features; ++i) {
|
||||||
|
if (i % 2 == 0) {
|
||||||
|
data[i] = std::floor(cat_d(&gen));
|
||||||
|
} else {
|
||||||
|
data[i] = num_d(&gen);
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
|
auto m = GetDMatrixFromData(data, n_samples, n_features);
|
||||||
|
m->Info().feature_types.HostVector().push_back(FeatureType::kCategorical);
|
||||||
|
m->Info().feature_types.HostVector().push_back(FeatureType::kNumerical);
|
||||||
|
|
||||||
|
auto cuts = DeviceSketch(0, m.get(), 64);
|
||||||
|
ASSERT_EQ(cuts.Values().size(), 64 + n_categories);
|
||||||
|
}
|
||||||
|
|
||||||
|
TEST(HistUtil, DeviceSketchMixedFeatures) {
|
||||||
|
TestMixedSketch();
|
||||||
|
}
|
||||||
|
|
||||||
TEST(HistUtil, DeviceSketchMultipleColumns) {
|
TEST(HistUtil, DeviceSketchMultipleColumns) {
|
||||||
int bin_sizes[] = {2, 16, 256, 512};
|
int bin_sizes[] = {2, 16, 256, 512};
|
||||||
int sizes[] = {100, 1000, 1500};
|
int sizes[] = {100, 1000, 1500};
|
||||||
|
|||||||
Loading…
x
Reference in New Issue
Block a user