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

Initial support for external memory in gradient index. (#7183)

Browse Source 
* Add hessian to batch param in preparation of new approx impl.
* Extract a push method for gradient index matrix.
* Use span instead of vector ref for hessian in sketching.
* Create a binary format for gradient index.
This commit is contained in:
Jiaming Yuan
2021-09-13 12:40:56 +08:00
committed by GitHub
parent a0dcf6f5c1
commit 3515931305
26 changed files with 546 additions and 171 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
src/data/simple_dmatrix.cc
View File

@@ -94,10 +94,12 @@ BatchSet<GHistIndexMatrix> SimpleDMatrix::GetGradientIndex(const BatchParam& par
if (!(batch_param_ != BatchParam{})) {
CHECK(param != BatchParam{}) << "Batch parameter is not initialized.";
}
if (!gradient_index_ || (batch_param_ != param && param != BatchParam{})) {
if (!gradient_index_ || (batch_param_ != param && param != BatchParam{}) || param.regen) {
CHECK_GE(param.max_bin, 2);
gradient_index_.reset(new GHistIndexMatrix(this, param.max_bin));
CHECK_EQ(param.gpu_id, -1);
gradient_index_.reset(new GHistIndexMatrix(this, param.max_bin, param.hess));
batch_param_ = param;
CHECK_EQ(batch_param_.hess.data(), param.hess.data());
}
auto begin_iter = BatchIterator<GHistIndexMatrix>(
new SimpleBatchIteratorImpl<GHistIndexMatrix>(gradient_index_));