MTUOC-paracrawl6.0-spa-eng / config-server.yaml
aoliverg's picture
Upload 39 files
9fa4f9e verified
raw
history blame
2.5 kB
MTEngine:
MTengine: Marian
#one of Marian, OpenNMT, Moses, GoogleTranslate, DeepL, Lucy
startMTEngine: True
startCommand: "./marian-server-GPU -m model.bin -v vocab-es.yml vocab-en.yml -p 8200 --n-best --alignment hard --normalize 1 --quiet &"
IP: localhost
port: 8200
min_len_factor: 0.5
MTUOCServer:
port: 8000
type: MTUOC
#one of MTUOC, Moses, ModernMT, OpenNMT, NMTWizard
verbosity_level: 3
checkistranslatable: True
restore_tags: True
fix_xml: True
URLs: True
code_URLs: "@URL@"
EMAILs: True
code_EMAILs: "@EMAIL@"
replaceNUMs: True
splitNUMs: False
min_chars_segment: 2
translation_selection_strategy: First
log_file: log.log
ONMT_url_root: "/translator"
SRX_file: segment.srx
Preprocess:
sl_lang: es
tl_lang: en
tokenize_SL: False
sl_tokenizer: MTUOC_tokenizer_spa.py
#one of the MTUOC-tokenizers or state Moses if you use Moses tokenizers
tl_tokenizer: MTUOC_tokenizer_eng.py
#one of the MTUOC-tokenizers or state Moses if you use Moses tokenizers
segment_input: True
SRXfile: segmentSC.srx
SRXlang: Spanish
replaceNUMs: False
code_NUMs: "@NUM@"
splitNUMs: False
#Truecasing
tcmodel: tc.es
#or None if not used
truecase: upper
#one of None, all, upper
###SentencePiece###
sentencepiece: True
sp_model_SL: spmodel.model
sp_model_TL: spmodel.model
sp_splitter: "▁"
#BPE
BPE: False
#True or False: if sentencepiece is set to True, sentencepiece will be used
bpecodes: codes_file
bpe_joiner: "@@"
bos_annotate: True
bos_symbol: <s>
#None or <s> (or other)
eos_annotate: True
eos_symbol: </s>
#hmtl entities
unescape_html_input: True
escape_html_input: False
unescape_html_output: True
escape_html_output: False
change_input_files: None
change_output_files: None
change_translation_files: None
GoogleTranslate:
sllang: en
tllang: es
glossary: None
#state None if no glossary is used, otherwise the name of the glossary
project_id: XXxxXXXXX
location: us-central1
jsonfile: XXxxXXXXX.json
DeepL:
API_key: XXxxXXXXX
sllang: EN
tllang: ES
#tag_handling: html
#one of html, xml
formality: default
#one of default, less, more
split_sentences: "off"
#one of "off", "on", "nonewlines"
glossary: None
Lucy:
url: http://XXX.XXX.XXX:8080/AutoTranslateRS/V1.3/mtrans/exec/
TRANSLATION_DIRECTION: SPANISH-CATALAN
MARK_UNKNOWNS: 0
MARK_ALTERNATIVES: 0
MARK_COMPOUNDS: 0
CHARSET: UTF