| 378 | } |
| 379 | |
| 380 | static void CreateFoldData( |
| 381 | typename TTrainingDataProviders::TDataPtr srcData, |
| 382 | ui64 cpuUsedRamLimit, |
| 383 | const TVector<NCB::TArraySubsetIndexing<ui32>>& trainSubsets, |
| 384 | const TVector<NCB::TArraySubsetIndexing<ui32>>& testSubsets, |
| 385 | TVector<TTrainingDataProviders>* foldsData, |
| 386 | TVector<TTrainingDataProviders>* testFoldsData, |
| 387 | NPar::ILocalExecutor* localExecutor |
| 388 | ) { |
| 389 | CB_ENSURE_INTERNAL(trainSubsets.size() == testSubsets.size(), "Number of train and test subsets do not match"); |
| 390 | const NCB::EObjectsOrder objectsOrder = NCB::EObjectsOrder::Ordered; |
| 391 | const ui64 perTaskCpuUsedRamLimit = cpuUsedRamLimit / (2 * trainSubsets.size()); |
| 392 | |
| 393 | TVector<std::function<void()>> tasks; |
| 394 | for (ui32 foldIdx : xrange(trainSubsets.size())) { |
| 395 | tasks.emplace_back( |
| 396 | [&, foldIdx]() { |
| 397 | (*foldsData)[foldIdx].Learn = srcData->GetSubset( |
| 398 | GetSubset( |
| 399 | srcData->ObjectsGrouping, |
| 400 | NCB::TArraySubsetIndexing<ui32>(trainSubsets[foldIdx]), |
| 401 | objectsOrder |
| 402 | ), |
| 403 | perTaskCpuUsedRamLimit, |
| 404 | localExecutor |
| 405 | ); |
| 406 | } |
| 407 | ); |
| 408 | tasks.emplace_back( |
| 409 | [&, foldIdx]() { |
| 410 | (*testFoldsData)[foldIdx].Test.emplace_back( |
| 411 | srcData->GetSubset( |
| 412 | GetSubset( |
| 413 | srcData->ObjectsGrouping, |
| 414 | NCB::TArraySubsetIndexing<ui32>(testSubsets[foldIdx]), |
| 415 | objectsOrder |
| 416 | ), |
| 417 | perTaskCpuUsedRamLimit, |
| 418 | localExecutor |
| 419 | ) |
| 420 | ); |
| 421 | } |
| 422 | ); |
| 423 | } |
| 424 | |
| 425 | NCB::ExecuteTasksInParallel(&tasks, localExecutor); |
| 426 | } |
| 427 | |
| 428 | static void TakeMiddleElements( |
| 429 | ui32 offset, |
no test coverage detected