Last active
November 9, 2016 15:57
-
-
Save naiaden/497afff85e205d47079f3ff73d5c4904 to your computer and use it in GitHub Desktop.
preprocessing
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| export W=2 | |
| export T=2 | |
| export t=2 | |
| export SAMPLES=10 | |
| export DOMAIN=$(dnsdomainname) | |
| if [ ${DOMAIN} = "science.ru.nl" ]; then | |
| OUTPUTPREFIX=/scratch/lonrust/cococpypv3 | |
| COCOCPYPPREFIX=/home/lonrust/Software/cococpyp | |
| else | |
| OUTPUTPREFIX=/esat/spchdisk/scratch/onrust/cococpypv3 | |
| COCOCPYPPREFIX=/users/spraak/onrust/software/cococpyp | |
| fi | |
| export OUTPUTDIR=${OUTPUTPREFIX}/output | |
| export INPUTDIR=${OUTPUTPREFIX}/input | |
| export DERIVEDDIR=${OUTPUTPREFIX}/derived | |
| export MODELDIR=${OUTPUTPREFIX}/models | |
| export COCOCPYPDIR=${COCOCPYPPREFIX}/hpyplm | |
| export TEST=${1:-jrc} | |
| export TRAIN=1bw | |
| export DEBUG=NONE | |
| export BACKOFF=${2:-uni} | |
| export PREFIX=${BACKOFF}-- | |
| export W=100 | |
| if [ ${BACKOFF} = "mle" ]; then | |
| LIMITEDCACHEFILE=limitedmlecachefile; | |
| elif [ ${BACKOFF} = "ent" ]; then | |
| LIMITEDCACHEFILE=limitedentropycachefile; | |
| else | |
| LIMITEDCACHEFILE=limiteduniformcachefile; | |
| fi | |
| ${COCOCPYPDIR}/hq4 -o ${MODELDIR} -m ${TRAIN}_4S_W${W}_t${t}_T${T}_s${SAMPLES}_p0_v2_train -M ${PREFIX}${TRAIN}_4S_W${W}_t${t}_T${T}_s${SAMPLES}_p0_v2-${TEST} -F ${INPUTDIR}/test-${TEST}.4gram-sm.txt -O ${OUTPUTDIR} -C ${DERIVEDDIR}/${TRAIN}_4S_W${W}_t${t}_T${T}_s${SAMPLES}_p0_v2_train.copamo --debug ${DEBUG} -B ${BACKOFF} -L ${OUTPUTDIR}/${PREFIX}${TRAIN}-4gramsm_4_W${W}_t${t}_T${T}_s${SAMPLES}_p0_v2-${TEST}_ngram-common_4.lim --${LIMITEDCACHEFILE} ${DERIVEDDIR}/${BACKOFF}.dummy | |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| W=2 | |
| T=2 | |
| t=2 | |
| SIZE=100000 | |
| SIZES=100k | |
| head -n ${SIZE} ~/data/cococpypv2/input/train-1bw.4gram-sm.txt > ~/data/cococpypv2/input/train-1bw.${SIZES}.4gram-sm.txt | |
| for i in 1 2 3 4 5; do | |
| colibri-freqlist -t ${t} -m ${i} -l ${i} -s ~/data/cococpypv2/input/train-1bw.${SIZES}.4gram-sm.txt > ~/data/cococpypv2/derived/train-1bw-${SIZES}-4gramsm-t2-s.${i}; | |
| done | |
| ~/Software/cococpyp/hpyplm/hpyplm_train4 -o ~/data/cococpypv2/models -m 1bw-4gramsm-${SIZES} -f ~/data/cococpypv2/input/train-1bw.${SIZES}.4gram-sm.txt -s 10 -T ${T} -t ${t} -W ${W} | |
| PPWD=$(pwd); cd ~/Software/cococpyp/hpyplm; rm hq4; make hq4; cd ${PPWD}; tput reset | |
| ~/Software/cococpyp/hpyplm/hq4 -o ~/data/cococpypv2/models -m 1bw-4gramsm-${SIZES}_4_W${W}_t${t}_T${T}_s10_p0_v2_train -M 1bw-4gramsm-${SIZES}_4_W${W}_t${t}_T${T}_s10_p0_v2-1bw -F ~/data/cococpypv2/input/test-1bw.4gram-sm.txt -O ~/data/cococpypv2/output -C ~/data/cococpypv2/derived/train-1bw-${SIZES}-4gramsm-t${t}-s -L ~/data/cococpypv2/output/1bw-4gramsm-100k_4_W2_t2_T2_s10_p0_v2-1bw_ngram-common_4.lim | |
| cut -d' ' -f4- fv800001.0.SPKR02-5194-6768.10000-best.txt | sed 's/^/<s> <s> /' > fv800001.0.SPKR02-5194-6768.10000-best.4gram-sm.txt | |
| colibri-patternmodeller -c /scratch/lonrust/cococpypv3/models/mediargus_4S_W250_t2_T2_s10_p0_v2_train.cls -f /scratch/lonrust/cococpypv3/models/mediargus_4S_W250_t2_T2_s10_p0_v2_train.dat -t 2 -m 1 -l 5 -W 250 -s -P > /scratch/lonrust/cococpypv3/output/mediargus_4S_W250_t2_T2_s10_p0_v2_train.copamo | |
| cut -f1,2 /scratch/lonrust/cococpypv3/output/mediargus_4S_W250_t2_T2_s10_p0_v2_train.copamo | tail -n +2 | sort -T /scratch/lonrust --parallel=16 -k2,2rn -k1,1 -t $'\t' > /scratch/lonrust/cococpypv3/output/mediargus_4S_W250_t2_T2_s10_p0_v2_train.copamo.sorted | |
| for i in 1 2 3 4 5; do grep -P "^[^ ]+( [^ ]+){$((i-1))}\t" /scratch/lonrust/cococpypv3/output/mediargus_4S_W250_t2_T2_s10_p0_v2_train.copamo.sorted > /scratch/lonrust/cococpypv3/output/mediargus_4S_W250_t2_T2_s10_p0_v2_train.copamo.$i; done | |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| export W=2 | |
| export T=2 | |
| export t=2 | |
| export SAMPLES=10 | |
| if [ $(dnsdomainname) = "science.ru.nl" ]; then | |
| OUTPUTPREFIX=/scratch/lonrust/cococpypv3 | |
| COCOCPYPPREFIX=/home/lonrust/Software/cococpyp | |
| else | |
| OUTPUTPREFIX=/esat/spchdisk/scratch/onrust/cococpypv3 | |
| COCOCPYPPREFIX=/users/spraak/onrust/software/cococpyp | |
| fi | |
| export OUTPUTDIR=${OUTPUTPREFIX}/output | |
| export INPUTDIR=${OUTPUTPREFIX}/input | |
| export DERIVEDDIR=${OUTPUTPREFIX}/derived | |
| export MODELDIR=${OUTPUTPREFIX}/models | |
| export COCOCPYPDIR=${COCOCPYPPREFIX}/hpyplm | |
| export TEST=jrc | |
| export TRAIN=1bw | |
| export DEBUG=NONE | |
| ONLYTEST=true | |
| function run_subset { | |
| export SIZE=100000 | |
| export SIZES=100k | |
| if [ ${ONLYTEST} = "false" ]; then | |
| head -n ${SIZE} ${INPUTDIR}/train-${TRAIN}.4gram-sm.txt > ${DERIVEDDIR}/train-${TRAIN}.${SIZES}.4gram-sm.txt | |
| parallel --env SIZES --env t --env DERIVEDDIR 'colibri-freqlist -t ${t} -m {1} -l {1} -s ${DERIVEDDIR}/train-${TRAIN}.${SIZES}.4gram-sm.txt > ${DERIVEDDIR}/train-${TRAIN}-${SIZES}-4gramsm-t${t}-s.{1}' ::: 1 2 3 4 5 | |
| PPWD=$(pwd); cd ${COCOCPYPDIR}; \rm hq4; make hq4; \rm ht4; make ht4; cd ${PPWD}; tput reset | |
| ${COCOCPYPDIR}/ht4 -o ${MODELDIR} -m ${TRAIN}-4gramsm-${SIZES} -f ${DERIVEDDIR}/train-${TRAIN}.${SIZES}.4gram-sm.txt -s ${SAMPLES} -T ${T} -t ${t} -W ${W} | |
| fi | |
| BACKOFF=uni | |
| PREFIX=${BACKOFF}-- | |
| ${COCOCPYPDIR}/hq4 -o ${MODELDIR} -m ${TRAIN}-4gramsm-${SIZES}_4_W${W}_t${t}_T${T}_s${SAMPLES}_p0_v2_train -M ${PREFIX}${TRAIN}-4gramsm-${SIZES}_4_W${W}_t${t}_T${T}_s${SAMPLES}_p0_v2-${TEST} -F ${INPUTDIR}/test-${TEST}.4gram-sm.txt -O ${OUTPUTDIR} -C ${DERIVEDDIR}/train-${TRAIN}-${SIZES}-4gramsm-t${t}-s -L ${OUTPUTDIR}/${TRAIN}-4gramsm-${SIZES}_4_W${W}_t${t}_T${T}_s${SAMPLES}_p0_v2-${TEST}_ngram-common_4.lim --debug ${DEBUG} -B ${BACKOFF} | |
| BACKOFF=mle | |
| PREFIX=${BACKOFF}-- | |
| ${COCOCPYPDIR}/hq4 -o ${MODELDIR} -m ${TRAIN}-4gramsm-${SIZES}_4_W${W}_t${t}_T${T}_s${SAMPLES}_p0_v2_train -M ${PREFIX}${TRAIN}-4gramsm-${SIZES}_4_W${W}_t${t}_T${T}_s${SAMPLES}_p0_v2-${TEST} -F ${INPUTDIR}/test-${TEST}.4gram-sm.txt -O ${OUTPUTDIR} -C ${DERIVEDDIR}/train-${TRAIN}-${SIZES}-4gramsm-t${t}-s -L ${OUTPUTDIR}/${TRAIN}-4gramsm-${SIZES}_4_W${W}_t${t}_T${T}_s${SAMPLES}_p0_v2-${TEST}_ngram-common_4.lim --backoff ${BACKOFF} --debug ${DEBUG} | |
| BACKOFF=ent | |
| PREFIX=${BACKOFF}-- | |
| ${COCOCPYPDIR}/hq4 -o ${MODELDIR} -m ${TRAIN}-4gramsm-${SIZES}_4_W${W}_t${t}_T${T}_s${SAMPLES}_p0_v2_train -M ${PREFIX}${TRAIN}-4gramsm-${SIZES}_4_W${W}_t${t}_T${T}_s${SAMPLES}_p0_v2-${TEST} -F ${INPUTDIR}/test-${TEST}.4gram-sm.txt -O ${OUTPUTDIR} -C ${DERIVEDDIR}/train-${TRAIN}-${SIZES}-4gramsm-t${t}-s -L ${OUTPUTDIR}/${TRAIN}-4gramsm-${SIZES}_4_W${W}_t${t}_T${T}_s${SAMPLES}_p0_v2-${TEST}_ngram-common_4.lim --backoff ${BACKOFF} --debug ${DEBUG} | |
| } | |
| function run_all { | |
| if [ ${ONLYTEST} = "false" ]; then | |
| #parallel --env SIZES --env t --env DERIVEDDIR 'colibri-freqlist -t ${t} -m {1} -l {1} -s ${INPUTDIR}/train-${TRAIN}.4gram-sm.txt > ${DERIVEDDIR}/train-${TRAIN}-4gramsm-t${t}-s.{1}' ::: 1 2 3 4 5 | |
| # 1bw_4S_W100_t2_T2_s10_p0_v2_train.copamo | |
| PPWD=$(pwd); cd ${COCOCPYPDIR}; \rm hq4; make hq4; \rm ht4; make ht4; cd ${PPWD}; tput reset | |
| ${COCOCPYPDIR}/ht4 -o ${MODELDIR} -m ${TRAIN}-4gramsm -f ${DERIVEDDIR}/train-${TRAIN}.4gram-sm.txt -s ${SAMPLES} -T ${T} -t ${t} -W ${W} | |
| fi | |
| export BACKOFF=uni | |
| export PREFIX=${BACKOFF}-- | |
| export W=100 | |
| ${COCOCPYPDIR}/hq4 -o ${MODELDIR} -m ${TRAIN}_4S_W${W}_t${t}_T${T}_s${SAMPLES}_p0_v2_train -M ${PREFIX}${TRAIN}_4S_W${W}_t${t}_T${T}_s${SAMPLES}_p0_v2-${TEST} -F ${INPUTDIR}/test-${TEST}.4gram-sm.txt -O ${OUTPUTDIR} -C ${DERIVEDDIR}/${TRAIN}_4S_W${W}_t${t}_T${T}_s${SAMPLES}_p0_v2_train.copamo --debug ${DEBUG} -B ${BACKOFF} -L ${OUTPUTDIR}/${PREFIX}${TRAIN}-4gramsm_4_W${W}_t${t}_T${T}_s${SAMPLES}_p0_v2-${TEST}_ngram-common_4.lim --limiteduniformcachefile ${DERIVEDDIR}/uni.dummy | |
| export BACKOFF=mle | |
| export PREFIX=${BACKOFF}-- | |
| export W=100 | |
| ${COCOCPYPDIR}/hq4 -o ${MODELDIR} -m ${TRAIN}_4S_W${W}_t${t}_T${T}_s${SAMPLES}_p0_v2_train -M ${PREFIX}${TRAIN}_4S_W${W}_t${t}_T${T}_s${SAMPLES}_p0_v2-${TEST} -F ${INPUTDIR}/test-${TEST}.4gram-sm.txt -O ${OUTPUTDIR} -C ${DERIVEDDIR}/${TRAIN}_4S_W${W}_t${t}_T${T}_s${SAMPLES}_p0_v2_train.copamo --debug ${DEBUG} -B ${BACKOFF} -L ${OUTPUTDIR}/${PREFIX}${TRAIN}-4gramsm_4_W${W}_t${t}_T${T}_s${SAMPLES}_p0_v2-${TEST}_ngram-common_4.lim --limitedmlecachefile ${DERIVEDDIR}/mle.dummy | |
| export BACKOFF=ent | |
| export PREFIX=${BACKOFF}-- | |
| export W=100 | |
| ${COCOCPYPDIR}/hq4 -o ${MODELDIR} -m ${TRAIN}_4S_W${W}_t${t}_T${T}_s${SAMPLES}_p0_v2_train -M ${PREFIX}${TRAIN}_4S_W${W}_t${t}_T${T}_s${SAMPLES}_p0_v2-${TEST} -F ${INPUTDIR}/test-${TEST}.4gram-sm.txt -O ${OUTPUTDIR} -C ${DERIVEDDIR}/${TRAIN}_4S_W${W}_t${t}_T${T}_s${SAMPLES}_p0_v2_train.copamo --debug ${DEBUG} -B ${BACKOFF} -L ${OUTPUTDIR}/${PREFIX}${TRAIN}-4gramsm_4_W${W}_t${t}_T${T}_s${SAMPLES}_p0_v2-${TEST}_ngram-common_4.lim --limitedentropycachefile ${DERIVEDDIR}/ent.dummy | |
| } | |
| SUBSET=true | |
| if [ ${SUBSET} = "true" ]; then | |
| rub_subset | |
| else | |
| run_all | |
| fi | |
| head -n ${SIZE} ${INPUTDIR}/train-${TRAIN}.4gram-sm.txt > ${DERIVEDDIR}/train-${TRAIN}.${SIZES}.4gram-sm.txt | |
| parallel --env SIZES --env t --env DERIVEDDIR 'colibri-freqlist -t ${t} -m {1} -l {1} -s ${DERIVEDDIR}/train-${TRAIN}.${SIZES}.4gram-sm.txt > ${DERIVEDDIR}/train-${TRAIN}-${SIZES}-4gramsm-t${t}-s.{1}' ::: 1 2 3 4 5 | |
| PPWD=$(pwd); cd ${COCOCPYPDIR}; \rm hq4; make hq4; \rm ht4; make ht4; cd ${PPWD}; tput reset | |
| ${COCOCPYPDIR}/ht4 -o ${MODELDIR} -m ${TRAIN}-4gramsm-${SIZES} -f ${DERIVEDDIR}/train-${TRAIN}.${SIZES}.4gram-sm.txt -s ${SAMPLES} -T ${T} -t ${t} -W ${W} | |
| BACKOFF=uni | |
| PREFIX=${BACKOFF}-- | |
| ${COCOCPYPDIR}/hq4 -o ${MODELDIR} -m ${TRAIN}-4gramsm-${SIZES}_4_W${W}_t${t}_T${T}_s${SAMPLES}_p0_v2_train -M ${PREFIX}${TRAIN}-4gramsm-${SIZES}_4_W${W}_t${t}_T${T}_s${SAMPLES}_p0_v2-${TEST} -F ${INPUTDIR}/test-${TEST}.4gram-sm.txt -O ${OUTPUTDIR} -C ${DERIVEDDIR}/train-${TRAIN}-${SIZES}-4gramsm-t${t}-s -L ${OUTPUTDIR}/${TRAIN}-4gramsm-${SIZES}_4_W${W}_t${t}_T${T}_s${SAMPLES}_p0_v2-${TEST}_ngram-common_4.lim --debug ${DEBUG} -B ${BACKOFF} | |
| BACKOFF=mle | |
| PREFIX=${BACKOFF}-- | |
| ${COCOCPYPDIR}/hq4 -o ${MODELDIR} -m ${TRAIN}-4gramsm-${SIZES}_4_W${W}_t${t}_T${T}_s${SAMPLES}_p0_v2_train -M ${PREFIX}${TRAIN}-4gramsm-${SIZES}_4_W${W}_t${t}_T${T}_s${SAMPLES}_p0_v2-${TEST} -F ${INPUTDIR}/test-${TEST}.4gram-sm.txt -O ${OUTPUTDIR} -C ${DERIVEDDIR}/train-${TRAIN}-${SIZES}-4gramsm-t${t}-s -L ${OUTPUTDIR}/${TRAIN}-4gramsm-${SIZES}_4_W${W}_t${t}_T${T}_s${SAMPLES}_p0_v2-${TEST}_ngram-common_4.lim --backoff ${BACKOFF} --debug ${DEBUG} | |
| BACKOFF=ent | |
| PREFIX=${BACKOFF}-- | |
| ${COCOCPYPDIR}/hq4 -o ${MODELDIR} -m ${TRAIN}-4gramsm-${SIZES}_4_W${W}_t${t}_T${T}_s${SAMPLES}_p0_v2_train -M ${PREFIX}${TRAIN}-4gramsm-${SIZES}_4_W${W}_t${t}_T${T}_s${SAMPLES}_p0_v2-${TEST} -F ${INPUTDIR}/test-${TEST}.4gram-sm.txt -O ${OUTPUTDIR} -C ${DERIVEDDIR}/train-${TRAIN}-${SIZES}-4gramsm-t${t}-s -L ${OUTPUTDIR}/${TRAIN}-4gramsm-${SIZES}_4_W${W}_t${t}_T${T}_s${SAMPLES}_p0_v2-${TEST}_ngram-common_4.lim --backoff ${BACKOFF} --debug ${DEBUG} |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment