mosesdecoder/contrib/m4m/modules/moses-parameters.m4m

58 lines
1.6 KiB
Makefile

# -*- makefile -*-
casing1 ?= truecase
casing2 ?= truecase
word-alignment ?= fast
moses.threads = $(shell parallel --number-of-cores)
# numerical constants for moses
moses.inputtype.plaintext = 0
PHRASE_TABLES = $(ptable)
LANGUAGE_MODELS = $(lmodel)
DISTORTION_MODELS = $(lexdm)
# defaults
ptable = model/tm/bin/${L1}-${L2}
lmodel = model/lm/${L2}/kenlm
lexdm_specs = wbe-mslr-bidirectional-fe-allff
lexdm = model/dm/bin/${L1}-${L2}/${dflt_lexdmodel_specs}
ptable.max-phrase-length ?= 7
ptable.smoothing ?= --GoodTuring
ptable.source-factors ?= 0
ptable.target-factors ?= 0
ptable.num-features ?= 5
ptable.implemetation ?= 1
# reminder: implementation types:
# 0 - text
# 1 - binary via processPhraseTable
# 14 - based on dynamic suffix arrays
# if you don't provide the phrase table, you must
# specify the components of the word-aligned corpus
# as target-specific variable assignments
model/tm/bin/${L1}-${L2}: ptable.L1_text = ${
model/tm/bin/${L1}-${L2}: ptable.smoothing = --GoodTuring
# source corpus and parameters here:
# reordering model
dmodel.type = wbe
dmodel.orientation = mslr
dmodel.direction = bidirectional
dmodel.language = fe
dmodel.collapsing = allff
dmodel.smooth = 0.5
dmodel.num-feats = 9
dmodel.description = $(addprefix ${dmodel.type}-${dmodel.orientation}-,\
${dmodel.direction}-${dmodel.language}-${dmodel.collapsing})
distortion-limit = 6
# DEFAULT TUNING PARAMETERS
FACTORSEP ?= \n
mert.nbest = 100
mert.extra-flags ?=
mert.extra-flags += --no-filter-phrase-table
mert.decoder-flags = -threads ${moses.threads} -fd '${FACTORSEP}'