From 8a77b1cb710bcb05f3eae5a4daf72ea11f8d22c8 Mon Sep 17 00:00:00 2001 From: Franck Dary <franck.dary@lis-lab.fr> Date: Thu, 7 May 2020 10:38:27 +0200 Subject: [PATCH] updated scripts to take into accound specifics mcd --- UD_any/evaluate.sh | 10 ++++++++-- UD_any/train.sh | 15 +++++++++++---- 2 files changed, 19 insertions(+), 6 deletions(-) diff --git a/UD_any/evaluate.sh b/UD_any/evaluate.sh index dc0869d..0a11dd9 100755 --- a/UD_any/evaluate.sh +++ b/UD_any/evaluate.sh @@ -40,15 +40,17 @@ DEV=$(find $CORPUS -type f -name '*dev*.conllu') DEVRAW=$(find $CORPUS -type f -name '*dev*.txt') TEST=$(find $CORPUS -type f -name '*test*.conllu') TESTRAW=$(find $CORPUS -type f -name '*test*.txt') +MCD=$(find $CORPUS -type f -name '*.mcd') REF=$TEST REFRAW=$TESTRAW -if has_space "$REF" || has_space "$REFRAW"; +if has_space "$REF" || has_space "$REFRAW" || has_space "$MCD"; then >&2 echo "ERROR : more than 1 match" >&2 echo "REF : " $REF >&2 echo "REFRAW : " $REFRAW + >&2 echo "MCD : " $MCD print_usage_and_exit fi @@ -65,7 +67,11 @@ then print_usage_and_exit fi -MCD=$EXPPATH"/data/*\.mcd" +if test -z $MCD; +then + MCD=$EXPPATH"/data/*\.mcd" +fi + EVALCONLL="../scripts/conll18_ud_eval.py" OUTPUT=$EXPPATH"/predicted_eval.tsv" diff --git a/UD_any/train.sh b/UD_any/train.sh index 3762e77..a611cad 100755 --- a/UD_any/train.sh +++ b/UD_any/train.sh @@ -43,17 +43,19 @@ DEV=$(find $CORPUS -type f -name '*dev*.conllu') DEVRAW=$(find $CORPUS -type f -name '*dev*.txt') TEST=$(find $CORPUS -type f -name '*test*.conllu') TESTRAW=$(find $CORPUS -type f -name '*test*.txt') +MCD=$(find $CORPUS -type f -name '*.mcd') -if has_space "$TRAIN" || has_space "$DEV" || has_space "$TEST"; +if has_space "$TRAIN" || has_space "$DEV" || has_space "$TEST" || has_space "$MCD"; then >&2 echo "ERROR : more than 1 match with keyword" $KEYWORD >&2 echo "TRAIN : " $TRAIN >&2 echo "DEV : " $DEV >&2 echo "TEST : " $TEST + >&2 echo "MCD : " $MCD print_usage_and_exit fi -if test ! -f $TRAIN; +if test -z $TRAIN; then >&2 echo "ERROR : no train file found in" $CORPUS >&2 echo "$TRAIN" @@ -61,7 +63,7 @@ then fi if [ "$MODE" = "txt" ]; then -if test ! -f $TRAINRAW; +if test -z $TRAINRAW; then >&2 echo "ERROR : no train file found in" $CORPUS >&2 echo "$TRAINRAW" @@ -69,7 +71,12 @@ then fi fi -MCD=$EXPPATH"/data/*\.mcd" +if test -z $MCD; +then + MCD=$EXPPATH"/data/*\.mcd" +fi + +>&2 echo "Using MCD :" $MCD if [ "$MODE" = "tsv" ]; then macaon train --model $EXPPATH --mcd $MCD --trainTSV $TRAIN --devTSV $DEV "$@" || exit 1 -- GitLab