Namespace marian::data

This file contains nearly all BERT-related code and adds BERT-functionality on top of existing classes like TansformerEncoder and Classifier.

Classes

• Struct BinaryShortlistGenerator::Header

• Struct Example

• Struct WordAlignment::Point

• Class Batch

• Template Class BatchGenerator

• Template Class BatchIterator

• Class BatchStats

• Class BertBatch

• Class BinaryShortlistGenerator

• Class Corpus

• Class CorpusBase

• Class CorpusBatch

• Class CorpusBatchGenerator

• Class CorpusIterator

• Class CorpusNBest

• Class CorpusSQLite

• Class DataBatch

• Class Dataset

• Template Class DatasetBase

• Class FakeShortlistGenerator

• Class Input

• Class LexicalShortlistGenerator

• Class LSHShortlist

• Class LSHShortlistGenerator

• Class MNISTData

• Class QuicksandShortlistGenerator

• Class RNGEngine

• Class SentenceTuple

• Class SentenceTupleImpl

• Class Shortlist

• Class ShortlistGenerator

• Class SubBatch

• Class TextInput

• Class TextIterator

• Class WordAlignment

Functions

• Function marian::data::ConvertSoftAlignToHardAlign

• Function marian::data::createShortlistGenerator

• Template Function marian::data::get

• Function marian::data::isBinaryShortlist

• Function marian::data::lineFromNbest

• Function marian::data::numFromNbest

• Function marian::data::SoftAlignToString

Typedefs

• Typedef marian::data::Data

• Typedef marian::data::ExampleIterator

• Typedef marian::data::Examples

• Typedef marian::data::Labels

• Typedef marian::data::MaskBatch

• Typedef marian::data::SentBatch

• Typedef marian::data::SoftAlignment

• Typedef marian::data::WordBatch

• Typedef marian::data::WordMask

Variables

• Variable marian::data::BINARY_SHORTLIST_MAGIC