ROOT::Experimental::Internal::ML Namespace Reference

Classes
class	RBatchLoader
	Building and loading the batches from loaded chunks in RChunkLoader. More...
class	RBatchSink
	Writes RFlat2DMatrix batches to disk letting RDataLoaderEngine::Save() stay agnostic to the output format. More...
class	RClusterLoader
	Loads TTree/RNTuple clusters from one or more RDataFrames into RFlat2DMatrix buffers for ML training and validation. More...
class	RClusterLoaderFunctor
	Functor invoked by RDataFrame::Foreach to fill one row of an RFlat2DMatrix. More...
struct	RClusterRange
	Describes a contiguous range of entries within a single RDataFrame, corresponding to one TTree/RNTuple cluster boundary. More...
struct	RColumnLayout
	Where one of the loader's columns lives inside a batch-tensor row and how it's shaped. More...
class	RDataLoaderEngine
	In this class, the processes of loading clusters (see RClusterLoader) and creating batches from those clusters (see RBatchLoader) are combined, allowing batches from the training and validation sets to be loaded directly from a dataset in an RDataFrame. More...
class	RDatasetLoader
	Load the whole dataset into memory. More...
class	RDatasetLoaderFunctor
	Loading chunks made in RDatasetLoader into tensors from data from RDataFrame. More...
struct	RFlat2DMatrix
	Wrapper around ROOT::RVec<float> representing a 2D matrix. More...
class	RFlat2DMatrixOperators
	Collection of operations applied to one or multiple flat 2D matrices. More...
class	RNTupleBatchSink
	Writes batches into an RNTuple, as one float field per scalar column and one std::vector<float> field per vector column. More...
class	RSampler
	Implementation of different sampling strategies. More...
class	RTTreeBatchSink
	Writes batches into a TTree, as one float leaf per scalar column and one std::vector<float> branch per vector column, matching the RNTuple sink's schema. More...

Functions
std::unique_ptr< RBatchSink >	CreateBatchSink (std::string_view dataset_name, std::string_view filename, std::vector< RColumnLayout > layout, std::string_view format)
	Create the sink matching format.

Function Documentation

CreateBatchSink()

std::unique_ptr< RBatchSink > ROOT::Experimental::Internal::ML::CreateBatchSink	(	std::string_view	dataset_name,
		std::string_view	filename,
		std::vector< RColumnLayout >	layout,
		std::string_view	format )

Create the sink matching format.

Parameters

dataset_name	Name of the output TTree or RNTuple.
filename	Path of the output file, overwritten if it exists.
layout	Where each of the loader's columns sits in a batch row, see RColumnLayout.
format	Either "ttree" or "rntuple".

Definition at line 116 of file RBatchSink.cxx.