Sequence modeling

Language models

Vocabulary	Stores language model vocabulary.
NgramCounter	Class for counting ngrams.
MLE	Class for providing MLE ngram model scores.
Lidstone	Provides Lidstone-smoothed scores.
Laplace	Implements Laplace (add one) smoothing.
WittenBellInterpolated	Interpolated version of Witten-Bell smoothing.
KneserNeyInterpolated	Interpolated version of Kneser-Ney smoothing.
AbsoluteDiscountingInterpolated	Interpolated version of smoothing with absolute discount.
StupidBackoff	Provides StupidBackoff scores.

Translation

AlignedSent	Return an aligned sentence object, which encapsulates two sentences along with an Alignment between them.
Alignment	A storage class for representing alignment between two sequences, s1, s2.
PhraseTable	In-memory store of translations for a given phrase, and the log probability of the those translations
IBMModel	Abstract base class for all IBM models
IBMModel1	Lexical translation model that ignores word order
IBMModel2	Lexical translation model that considers word order
IBMModel3	Translation model that considers how a word can be aligned to multiple words in another language
IBMModel4	Translation model that reorders output words based on their type and their distance from other related words in the output sentence
IBMModel5	Translation model that keeps track of vacant positions in the target sentence to decide where to place translated words
StackDecoder	Phrase-based stack decoder for machine translation
trace(backlinks, source_sents_lens, ...)	Traverse the alignment cost from the tracebacks and retrieves appropriate sentence pairs.
grow_diag_final_and(srclen, trglen, e2f, f2e)	This module symmetrisatizes the source-to-target and target-to-source word alignment output and produces, aka.
extract(f_start, f_end, e_start, e_end, ...)	This function checks for alignment point consistency and extracts phrases using the chunk of consistent phrases.