moses-smt/mosesdecoder
1
1
Fork 0
mirror of https://github.com/moses-smt/mosesdecoder.git synced 2024-09-11 11:25:40 +03:00
Code Issues Packages Projects Releases Wiki Activity

Interpolated OSM - Bug Fix

Browse Source
This commit is contained in:
Nadir 2015-10-08 12:03:09 +01:00
parent 2ec6fed898
commit 965aeb9012
5 changed files with 100 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

22
scripts/ems/example/config.basic
View File

@ -391,6 +391,28 @@ alignment-symmetrization-method = grow-diag-final-and
#operation-sequence-model-order = 5
#operation-sequence-model-settings = "-lmplz '$moses-src-dir/bin/lmplz -S 40% -T $working-dir/model/tmp'"
#
# OR if you want to use with SRILM
#
#operation-sequence-model-settings = "--srilm-dir /path-to-srilm/bin/i686-m64"
## Class-based Operation Sequence Model (OSM)
# if OSM has to be enabled with factors then add factors as below.
# Durrani, Koehn, Schmid, Fraser (COLING, 2014).
#Investigating the Usefulness of Generalized Word Representations in SMT
#
#operation-sequence-model-settings = "--factor 0-0+1-1"
## Interpolated Operation Sequence Model (OSM)
# if OSM has to be enabled with factors then add factors as below.
# Durrani, Sajjad, Joty, Abdelali and Vogel (Mt Summit, 2015).
# Using Joint Models for Domain Adaptation in Statistical Machine Translation
#
#interpolated-operation-sequence-model = "yes"
#operation-sequence-model-order = 5
#operation-sequence-model-settings = "--srilm-dir /path-to-srilm/bin/i686-m64 --tune /path-to-tune-folder/tune_file"
#Interpolated OSM can only be used with SRILM because of the interpolation script
# if OSM training should be skipped, point to OSM Model
#osm-model =

23
scripts/ems/example/config.factored
View File

@ -411,9 +411,30 @@ alignment-symmetrization-method = grow-diag-final-and
#operation-sequence-model-order = 5
#operation-sequence-model-settings = "-lmplz '$moses-src-dir/bin/lmplz -S 40% -T $working-dir/model/tmp'"
#
# OR if you want to use with SRILM
#
#operation-sequence-model-settings = "--srilm-dir /path-to-srilm/bin/i686-m64"
## Class-based Operation Sequence Model (OSM)
# if OSM has to be enabled with factors then add factors as below.
# Durrani, Koehn, Schmid, Fraser (COLING, 2014).
#Investigating the Usefulness of Generalized Word Representations in SMT
#
#operation-sequence-model-settings = "--factor 0-0+1-1"
## Interpolated Operation Sequence Model (OSM)
# if OSM has to be enabled with factors then add factors as below.
# Durrani, Sajjad, Joty, Abdelali and Vogel (Mt Summit, 2015).
# Using Joint Models for Domain Adaptation in Statistical Machine Translation
#
#interpolated-operation-sequence-model = "yes"
#operation-sequence-model-order = 5
#operation-sequence-model-settings = "--srilm-dir /path-to-srilm/bin/i686-m64 --tune /path-to-tune-folder/tune_file"
#Interpolated OSM can only be used with SRILM because of the interpolation script
# if OSM training should be skipped, point to OSM Model
#osm-model =
### unsupervised transliteration module
# Durrani, Sajjad, Hoang and Koehn (EACL, 2014).
# "Integrating an Unsupervised Transliteration Model

24
scripts/ems/example/config.toy
View File

@ -373,8 +373,30 @@ alignment-symmetrization-method = grow-diag-final-and
#
#operation-sequence-model = "yes"
#operation-sequence-model-order = 5
#operation-sequence-model-settings = ""
#operation-sequence-model-settings = "-lmplz '$moses-src-dir/bin/lmplz -S 40% -T $working-dir/model/tmp'"
#
# OR if you want to use with SRILM
#
#operation-sequence-model-settings = "--srilm-dir /path-to-srilm/bin/i686-m64"
## Class-based Operation Sequence Model (OSM)
# if OSM has to be enabled with factors then add factors as below.
# Durrani, Koehn, Schmid, Fraser (COLING, 2014).
#Investigating the Usefulness of Generalized Word Representations in SMT
#
#operation-sequence-model-settings = "--factor 0-0+1-1"
## Interpolated Operation Sequence Model (OSM)
# if OSM has to be enabled with factors then add factors as below.
# Durrani, Sajjad, Joty, Abdelali and Vogel (Mt Summit, 2015).
# Using Joint Models for Domain Adaptation in Statistical Machine Translation
#
#interpolated-operation-sequence-model = "yes"
#operation-sequence-model-order = 5
#operation-sequence-model-settings = "--srilm-dir /path-to-srilm/bin/i686-m64 --tune /path-to-tune-folder/tune_file"
#Interpolated OSM can only be used with SRILM because of the interpolation script
# if OSM training should be skipped, point to OSM Model
#osm-model =

24
scripts/ems/example/config.toy.bilinguallm
View File

@ -389,8 +389,30 @@ alignment-symmetrization-method = grow-diag-final-and
#
#operation-sequence-model = "yes"
#operation-sequence-model-order = 5
#operation-sequence-model-settings = ""
#operation-sequence-model-settings = "-lmplz '$moses-src-dir/bin/lmplz -S 40% -T $working-dir/model/tmp'"
#
# OR if you want to use with SRILM
#
#operation-sequence-model-settings = "--srilm-dir /path-to-srilm/bin/i686-m64"
## Class-based Operation Sequence Model (OSM)
# if OSM has to be enabled with factors then add factors as below.
# Durrani, Koehn, Schmid, Fraser (COLING, 2014).
#Investigating the Usefulness of Generalized Word Representations in SMT
#
#operation-sequence-model-settings = "--factor 0-0+1-1"
## Interpolated Operation Sequence Model (OSM)
# if OSM has to be enabled with factors then add factors as below.
# Durrani, Sajjad, Joty, Abdelali and Vogel (Mt Summit, 2015).
# Using Joint Models for Domain Adaptation in Statistical Machine Translation
#
#interpolated-operation-sequence-model = "yes"
#operation-sequence-model-order = 5
#operation-sequence-model-settings = "--srilm-dir /path-to-srilm/bin/i686-m64 --tune /path-to-tune-folder/tune_file"
#Interpolated OSM can only be used with SRILM because of the interpolation script
# if OSM training should be skipped, point to OSM Model
#osm-model =

12
scripts/ems/experiment.meta
View File

@ -533,7 +533,7 @@ build-domains
in: CORPUS:post-split-factorized-stem
out: domains
default-name: model/domains
ignore-unless: domain-features mml-filter-corpora operation-sequence-model
ignore-unless: domain-features mml-filter-corpora interpolated-operation-sequence-model
template: $moses-script-dir/ems/support/build-domain-file-from-subcorpora.perl $input-extension IN > OUT
final-model: yes
mml-score
@ -698,12 +698,20 @@ pcfg-score
pass-unless: use-pcfg-feature
template: ln -s IN.$input-extension OUT.$input-extension ; $moses-bin-dir/pcfg-score IN1.$output-extension < IN.$output-extension > OUT.$output-extension
build-osm
in: corpus word-alignment domains
in: corpus word-alignment
out: osm-model
ignore-unless: operation-sequence-model
rerun-on-change: operation-sequence-model training-options script giza-settings operation-sequence-model-settings
template: $moses-script-dir/OSM/OSM-Train.perl --corpus-f IN0.$input-extension --corpus-e IN0.$output-extension --alignment IN1.$alignment-symmetrization-method --order $operation-sequence-model-order --out-dir OUT --moses-src-dir $moses-src-dir --input-extension $input-extension --output-extension $output-extension $operation-sequence-model-settings
default-name: model/OSM
build-interpolated-osm
in: corpus word-alignment domains
out: osm-model
ignore-unless: interpolated-operation-sequence-model
rerun-on-change: interpolated-operation-sequence-model training-options script giza-settings operation-sequence-model-settings
template: $moses-script-dir/OSM/OSM-Train.perl --corpus-f IN0.$input-extension --corpus-e IN0.$output-extension --alignment IN1.$alignment-symmetrization-method --order $operation-sequence-model-order --out-dir OUT --moses-src-dir $moses-src-dir --input-extension $input-extension --output-extension $output-extension $operation-sequence-model-settings --domain IN2
default-name: model/OSM
build-transliteration-model
in: corpus word-alignment
out: transliteration-model