blob: 0d928af34fdb80e88312e788d2ab101eb28375e5 (
plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
|
#!/bin/bash -x
#####################################################################
# SUMMARY: Check if the ChrF validation measure equals to the SacreBLEU score
# AUTHOR: snukky
# TAGS: sentencepiece chrf valid-metrics
#####################################################################
# Exit on error
set -e
# Remove old artifacts and create working directory
rm -rf chrf chrf.*{log,out,diff,score}
mkdir -p chrf
# Copy the model
cp -r $MRT_MODELS/rnn-spm/model.npz chrf/
test -e chrf/model.npz
# Run marian command
$MRT_MARIAN/marian \
--no-shuffle --after-batches 1 --maxi-batch 1 --learn-rate 0 --overwrite \
-m chrf/model.npz -t $MRT_DATA/europarl.de-en/corpus.small.{de,en}.gz -v $MRT_MODELS/rnn-spm/vocab.{deen,deen}.spm \
--valid-freq 1 --valid-metrics chrf --valid-sets dev.de dev.en --valid-translation-output chrf.out \
--beam-size 8 --normalize 1 \
--log chrf.log
# Check if files exist
test -e chrf/model.npz
test -e chrf.out
test -e chrf.log
# Extract the score from logs
cat chrf.log | grep ' : chrf : ' | sed -r 's/.* chrf : (.*) : new best.*/\1/' > chrf.score
# Check score from logs
$MRT_TOOLS/diff.sh chrf.score chrf.score.expected > chrf.score.diff
# Run sacreBLEU removing the version information
python3 $MRT_TOOLS/sacrebleu/sacrebleu.py dev.en < chrf.out | sed -r 's/.version[^ ]* / /' > chrf.sacrebleu
# Check BLEU from the validation translation output
$MRT_TOOLS/diff.sh chrf.sacrebleu chrf.sacrebleu.expected > chrf.sacrebleu.diff
# Exit with success code
exit 0
|
