Helsinki-NLP/OPUS-MT-train
1
1
Fork 0
mirror of https://github.com/Helsinki-NLP/OPUS-MT-train.git synced 2024-07-14 23:50:23 +03:00
Code Issues Packages Projects Releases Wiki Activity

changed gitraw

Browse Source
This commit is contained in:
Joerg Tiedemann 2022-03-14 18:29:44 +02:00
parent 70c2d32913
commit f163518422
4 changed files with 46 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
lib/projects/distill.mk
View File

@ -105,6 +105,9 @@ test-quantized-all-finetuned-student:
%-small-student:
${MAKE} MODELTYPE=transformer-small-align ${@:small-student=student}
%-base-student:
${MAKE} MODELTYPE=transformer-base-align ${@:base-student=student}

40
lib/projects/elg.mk
View File

@ -66,11 +66,49 @@ engukr-quantize-student:
## special thing: student models with pivot-based data (does that work?)
## --> does not work very well ...
elg-ukr-students:
for l in bul dan deu fin hun nob ron swe slk tur; do \
${MAKE} STUDENT_DATA=pft-nopar SRCLANGS=ukr TRGLANGS=$$l train-tiny11-student; \
${MAKE} STUDENT_DATA=ftmono-pft-nopar SRCLANGS=ukr TRGLANGS=$$l train-tiny11-student; \
done
## tiny11 transformer model for finnish with pivot data (reuse student recipes)
elg-fin2ukr-tiny11:
${MAKE} STUDENT_DATA=pft-pbt SRCLANGS=fin TRGLANGS=ukr MARIAN_EXTRA=--no-restore-corpus train-tiny11-student
elg-ukr2fin-tiny11:
${MAKE} STUDENT_DATA=pft-pbt SRCLANGS=ukr TRGLANGS=fin train-tiny11-student
## tiny11 transformer model for finnish with pivot data (reuse student recipes)
elg-hun2ukr-tiny11:
${MAKE} STUDENT_DATA=pft-pbt-bt SRCLANGS=hun TRGLANGS=ukr MARIAN_EXTRA=--no-restore-corpus train-tiny11-student
elg-ukr2hun-tiny11:
${MAKE} STUDENT_DATA=pft-pbt-bt SRCLANGS=ukr TRGLANGS=hun train-tiny11-student
elg-ron2ukr-tiny11:
${MAKE} STUDENT_DATA=pft-pbt-bt SRCLANGS=ron TRGLANGS=ukr train-tiny11-student
elg-ukr2ron-tiny11:
${MAKE} STUDENT_DATA=pft-pbt-bt SRCLANGS=ukr TRGLANGS=ron train-tiny11-student
elg-swe2ukr-tiny11:
${MAKE} STUDENT_DATA=pft-pbt-bt SRCLANGS=swe TRGLANGS=ukr train-tiny11-student
elg-ukr2swe-tiny11:
${MAKE} STUDENT_DATA=pft-pbt-bt SRCLANGS=ukr TRGLANGS=swe train-tiny11-student
elg-ukr-students-test:
${MAKE} STUDENT_DATA=ftmono-pft-nopar SRCLANGS=ukr TRGLANGS=deu train-tiny11-student
${MAKE} STUDENT_DATA=ftmono-pft-nopar SRCLANGS=ukr TRGLANGS=hun train-tiny11-student

3
tatoeba/back-translate/Makefile
View File

@ -23,7 +23,8 @@ TRG = eng
TATOEBA_RELEASE = v2020-07-28
TATOEBA_STORAGE = https://object.pouta.csc.fi/Tatoeba-Challenge-${TATOEBA_RELEASE}
TATOEBA_WIKI_STORAGE = https://object.pouta.csc.fi/Tatoeba-Challenge-WikiShuffled
TATOEBA_GITRAW = https://raw.githubusercontent.com/Helsinki-NLP/Tatoeba-Challenge/master
# TATOEBA_GITRAW = https://raw.githubusercontent.com/Helsinki-NLP/Tatoeba-Challenge/master
TATOEBA_GITRAW = https://github.com/Helsinki-NLP/Tatoeba-Challenge/raw
TATOEBA_RELEASED = ${TATOEBA_GITRAW}/models/released-model-results.txt
TATOEBA_RELEASED_ALL = ${TATOEBA_GITRAW}/models/released-model-results-all.txt
TATOEBA_RELEASED_BT = https://object.pouta.csc.fi/Tatoeba-MT-bt/released-data.txt

3
tatoeba/pivoting/Makefile
View File

@ -30,7 +30,8 @@ MARIAN_WORKSPACE=12000
TATOEBA_VERSION ?= v2021-08-07
TATOEBA_VERSION_NOHYPHEN ?= $(subst -,,${TATOEBA_VERSION})
TATOEBA_GITRAW = https://raw.githubusercontent.com/Helsinki-NLP/Tatoeba-Challenge/master
# TATOEBA_GITRAW = https://raw.githubusercontent.com/Helsinki-NLP/Tatoeba-Challenge/master
TATOEBA_GITRAW = https://github.com/Helsinki-NLP/Tatoeba-Challenge/raw
TATOEBA_RELEASED = ${TATOEBA_GITRAW}/models/released-model-results-all.txt
TATOEBA_RELEASED_BT = https://object.pouta.csc.fi/Tatoeba-MT-bt/released-data.txt
TATOEBA_MODEL_STORAGE = https://object.pouta.csc.fi/Tatoeba-MT-models