mirror of
https://github.com/Helsinki-NLP/OPUS-MT-train.git
synced 2024-08-17 16:20:50 +03:00
makefile update
This commit is contained in:
Joerg Tiedemann
parent
106b06aa4c
commit
b52bbb676a
@ -76,16 +76,19 @@ BT_MODEL_BASE = ${BT_MODEL}.${MODELTYPE}.model${NR}
|
||||
BT_MODEL_START = ${WORKDIR}/${BT_MODEL_BASE}.npz
|
||||
BT_MODEL_VOCAB = ${WORKDIR}/${BT_MODEL}.vocab.${MODEL_VOCABTYPE}
|
||||
|
||||
%-add-backtranslations:
|
||||
# %-add-backtranslations:
|
||||
%-bt:
|
||||
ifneq (${wildcard ${MODEL_FINAL}},)
|
||||
ifeq (${wildcard ${BT_MODEL_START}},)
|
||||
cp ${MODEL_FINAL} ${BT_MODEL_START}
|
||||
cp ${MODEL_VOCAB} ${BT_MODEL_VOCAB}
|
||||
endif
|
||||
endif
|
||||
${MAKE} DATASET=opus+bt \
|
||||
CLEAN_TRAIN_SRC="${CLEAN_TRAIN_SRC} ${BACKTRANS_SRC}" \
|
||||
CLEAN_TRAIN_TRG="${CLEAN_TRAIN_TRG} ${BACKTRANS_TRG}" \
|
||||
MARIAN_EARLY_STOPPING=15 \
|
||||
${@:-add-backtranslations=}
|
||||
${@:-bt=}
|
||||
|
||||
|
||||
|
||||
|
||||
119
Makefile.tasks
119
Makefile.tasks
@ -53,6 +53,10 @@ multilingual:
|
||||
${MAKE} SRCLANGS="${LANGS}" TRGLANGS="${LANGS}" \
|
||||
WALLTIME=72 HPC_CORES=1 HPC_MEM=4g train.submit-multigpu
|
||||
|
||||
multilingual-big:
|
||||
${MAKE} SRCLANGS="${LANGS}" TRGLANGS="${LANGS}" data
|
||||
${MAKE} SRCLANGS="${LANGS}" TRGLANGS="${LANGS}" \
|
||||
WALLTIME=72 HPC_CORES=1 HPC_MEM=8g train.submit-multigpu
|
||||
|
||||
multilingual-medium:
|
||||
${MAKE} SRCLANGS="${LANGS}" TRGLANGS="${LANGS}" data
|
||||
@ -69,28 +73,24 @@ all2pivot:
|
||||
${MAKE} SRCLANGS="${PIVOT}" TRGLANGS="$$l" HPC_CORES=1 HPC_MEM=4g train.submit-multigpu; \
|
||||
done
|
||||
|
||||
bilingual-dynamic:
|
||||
|
||||
|
||||
## submit train jobs with settings that depend on the size of the training data
|
||||
## --> change WORKSPACE, MEM, nr of GPUs, validation frequency, stopping criterion
|
||||
|
||||
train-dynamic:
|
||||
if [ ! -e "${WORKHOME}/${LANGSTR}/train.submit" ]; then \
|
||||
${MAKE} data; \
|
||||
if [ `zcat ${WORKHOME}/${LANGSTR}/train/*.src.clean.${PRE_SRC}.gz | wc -l` -gt 10000000 ]; then \
|
||||
s=`zcat ${TRAIN_SRC}.clean.${PRE_SRC}${TRAINSIZE}.gz | head -10000001 | wc -l`; \
|
||||
if [ $$s -gt 10000000 ]; then \
|
||||
echo "${LANGSTR} bigger than 10 million"; \
|
||||
${MAKE} HPC_CORES=1 HPC_MEM=8g train.submit-multigpu; \
|
||||
if [ "${SRCLANGS}" != "${TRGLANGS}" ]; then \
|
||||
${MAKE} reverse-data-spm; \
|
||||
${MAKE} TRGLANGS="${SRCLANGS}" SRCLANGS='${TRGLANGS}' HPC_CORES=1 HPC_MEM=8g train.submit-multigpu; \
|
||||
fi; \
|
||||
elif [ `zcat ${WORKHOME}/${LANGSTR}/train/*.src.clean.${PRE_SRC}.gz | wc -l` -gt 1000000 ]; then \
|
||||
elif [ $$s -gt 1000000 ]; then \
|
||||
echo "${LANGSTR} bigger than 1 million"; \
|
||||
${MAKE} \
|
||||
MARIAN_VALID_FREQ=2500 \
|
||||
HPC_CORES=1 HPC_MEM=4g train.submit; \
|
||||
if [ "${SRCLANGS}" != "${TRGLANGS}" ]; then \
|
||||
${MAKE} reverse-data-spm; \
|
||||
${MAKE} TRGLANGS="${SRCLANGS}" SRCLANGS='${TRGLANGS}' \
|
||||
MARIAN_VALID_FREQ=2500 \
|
||||
HPC_CORES=1 HPC_MEM=4g train.submit; \
|
||||
fi; \
|
||||
elif [ `zcat ${WORKHOME}/${LANGSTR}/train/*.src.clean.${PRE_SRC}.gz | wc -l` -gt 100000 ]; then \
|
||||
elif [ $$s -gt 100000 ]; then \
|
||||
echo "${LANGSTR} bigger than 100k"; \
|
||||
${MAKE} \
|
||||
MARIAN_VALID_FREQ=1000 \
|
||||
@ -98,16 +98,7 @@ bilingual-dynamic:
|
||||
MARIAN_VALID_MINI_BATCH=8 \
|
||||
MARIAN_EARLY_STOPPING=5 \
|
||||
HPC_CORES=1 HPC_MEM=4g train.submit; \
|
||||
if [ "${SRCLANGS}" != "${TRGLANGS}" ]; then \
|
||||
${MAKE} reverse-data-spm; \
|
||||
${MAKE} TRGLANGS="${SRCLANGS}" SRCLANGS='${TRGLANGS}' \
|
||||
MARIAN_VALID_FREQ=1000 \
|
||||
MARIAN_WORKSPACE=5000 \
|
||||
MARIAN_VALID_MINI_BATCH=8 \
|
||||
MARIAN_EARLY_STOPPING=5 \
|
||||
HPC_CORES=1 HPC_MEM=4g train.submit; \
|
||||
fi; \
|
||||
elif [ `zcat ${WORKHOME}/${LANGSTR}/train/*.src.clean.${PRE_SRC}.gz | wc -l` -gt 10000 ]; then \
|
||||
elif [ $$s -gt 10000 ]; then \
|
||||
echo "${LANGSTR} bigger than 10k"; \
|
||||
${MAKE} \
|
||||
MARIAN_WORKSPACE=3500 \
|
||||
@ -116,22 +107,82 @@ bilingual-dynamic:
|
||||
MARIAN_VALID_FREQ=1000 \
|
||||
MARIAN_EARLY_STOPPING=5 \
|
||||
HPC_CORES=1 HPC_MEM=4g train.submit; \
|
||||
if [ "${SRCLANGS}" != "${TRGLANGS}" ]; then \
|
||||
${MAKE} reverse-data-spm; \
|
||||
${MAKE} TRGLANGS="${SRCLANGS}" SRCLANGS='${TRGLANGS}' \
|
||||
MARIAN_WORKSPACE=3500 \
|
||||
MARIAN_VALID_MINI_BATCH=4 \
|
||||
MARIAN_DROPOUT=0.5 \
|
||||
MARIAN_VALID_FREQ=1000 \
|
||||
MARIAN_EARLY_STOPPING=5 \
|
||||
HPC_CORES=1 HPC_MEM=4g train.submit; \
|
||||
fi; \
|
||||
else \
|
||||
echo "${LANGSTR} too small"; \
|
||||
fi \
|
||||
fi
|
||||
|
||||
|
||||
bilingual-dynamic: train-dynamic
|
||||
if [ "${SRCLANGS}" != "${TRGLANGS}" ]; then \
|
||||
${MAKE} reverse-data; \
|
||||
${MAKE} TRGLANGS="${SRCLANGS}" SRCLANGS='${TRGLANGS}' train-dynamic; \
|
||||
fi
|
||||
|
||||
# bilingual-dynamic:
|
||||
# if [ ! -e "${WORKHOME}/${LANGSTR}/train.submit" ]; then \
|
||||
# ${MAKE} data; \
|
||||
# s=`zcat ${WORKHOME}/${LANGSTR}/train/*.src.clean.${PRE_SRC}.gz | head -10000001 | wc -l`; \
|
||||
# if [ $$s -gt 10000000 ]; then \
|
||||
# echo "${LANGSTR} bigger than 10 million"; \
|
||||
# ${MAKE} HPC_CORES=1 HPC_MEM=8g train.submit-multigpu; \
|
||||
# if [ "${SRCLANGS}" != "${TRGLANGS}" ]; then \
|
||||
# ${MAKE} reverse-data-spm; \
|
||||
# ${MAKE} TRGLANGS="${SRCLANGS}" SRCLANGS='${TRGLANGS}' HPC_CORES=1 HPC_MEM=8g train.submit-multigpu; \
|
||||
# fi; \
|
||||
# elif [ $$s -gt 1000000 ]; then \
|
||||
# echo "${LANGSTR} bigger than 1 million"; \
|
||||
# ${MAKE} \
|
||||
# MARIAN_VALID_FREQ=2500 \
|
||||
# HPC_CORES=1 HPC_MEM=4g train.submit; \
|
||||
# if [ "${SRCLANGS}" != "${TRGLANGS}" ]; then \
|
||||
# ${MAKE} reverse-data-spm; \
|
||||
# ${MAKE} TRGLANGS="${SRCLANGS}" SRCLANGS='${TRGLANGS}' \
|
||||
# MARIAN_VALID_FREQ=2500 \
|
||||
# HPC_CORES=1 HPC_MEM=4g train.submit; \
|
||||
# fi; \
|
||||
# elif [ $$s -gt 100000 ]; then \
|
||||
# echo "${LANGSTR} bigger than 100k"; \
|
||||
# ${MAKE} \
|
||||
# MARIAN_VALID_FREQ=1000 \
|
||||
# MARIAN_WORKSPACE=5000 \
|
||||
# MARIAN_VALID_MINI_BATCH=8 \
|
||||
# MARIAN_EARLY_STOPPING=5 \
|
||||
# HPC_CORES=1 HPC_MEM=4g train.submit; \
|
||||
# if [ "${SRCLANGS}" != "${TRGLANGS}" ]; then \
|
||||
# ${MAKE} reverse-data-spm; \
|
||||
# ${MAKE} TRGLANGS="${SRCLANGS}" SRCLANGS='${TRGLANGS}' \
|
||||
# MARIAN_VALID_FREQ=1000 \
|
||||
# MARIAN_WORKSPACE=5000 \
|
||||
# MARIAN_VALID_MINI_BATCH=8 \
|
||||
# MARIAN_EARLY_STOPPING=5 \
|
||||
# HPC_CORES=1 HPC_MEM=4g train.submit; \
|
||||
# fi; \
|
||||
# elif [ $$s -gt 10000 ]; then \
|
||||
# echo "${LANGSTR} bigger than 10k"; \
|
||||
# ${MAKE} \
|
||||
# MARIAN_WORKSPACE=3500 \
|
||||
# MARIAN_VALID_MINI_BATCH=4 \
|
||||
# MARIAN_DROPOUT=0.5 \
|
||||
# MARIAN_VALID_FREQ=1000 \
|
||||
# MARIAN_EARLY_STOPPING=5 \
|
||||
# HPC_CORES=1 HPC_MEM=4g train.submit; \
|
||||
# if [ "${SRCLANGS}" != "${TRGLANGS}" ]; then \
|
||||
# ${MAKE} reverse-data-spm; \
|
||||
# ${MAKE} TRGLANGS="${SRCLANGS}" SRCLANGS='${TRGLANGS}' \
|
||||
# MARIAN_WORKSPACE=3500 \
|
||||
# MARIAN_VALID_MINI_BATCH=4 \
|
||||
# MARIAN_DROPOUT=0.5 \
|
||||
# MARIAN_VALID_FREQ=1000 \
|
||||
# MARIAN_EARLY_STOPPING=5 \
|
||||
# HPC_CORES=1 HPC_MEM=4g train.submit; \
|
||||
# fi; \
|
||||
# else \
|
||||
# echo "${LANGSTR} too small"; \
|
||||
# fi \
|
||||
# fi
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
Loading…
Reference in New Issue
Block a user