blob: 79cd801768ffedb6a260905d6a13c9e8402cbcbf (
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
|
#!/bin/bash -x
#####################################################################
# SUMMARY: Create a joint SentencePiece vocabulary
# AUTHOR: snukky
# TAGS: sentencepiece
#####################################################################
# Exit on error
set -e
# Remove old artifacts and create working directory
rm -rf vocab.joint vocab.joint.*{log,out,diff}
mkdir -p vocab.joint
# Run marian command
$MRT_MARIAN/marian \
--no-shuffle --seed 1111 --dim-emb 32 --dim-rnn 64 --maxi-batch 1 --maxi-batch-sort none --after-batches 1 \
-m vocab.joint/model.npz -t $MRT_DATA/europarl.de-en/corpus.small.{en,de}.gz \
--dim-vocabs 8000 -v vocab.joint/vocab.ende.spm vocab.joint/vocab.ende.spm --sentencepiece-options "--num_threads=1" \
--log vocab.joint.log
# Check if files exist
test -e vocab.joint/model.npz
test -e vocab.joint/vocab.ende.spm
test -e vocab.joint.log
# Check logging messages
grep -q "Training SentencePiece vocabulary .*vocab.ende.spm" vocab.joint.log
grep -q "Setting vocabulary size .* to 8,\?000" vocab.joint.log
grep -q "Sampling .* from .*corpus.small.de.gz, .*corpus.small.en.gz" vocab.joint.log
grep -q "Loading SentencePiece vocabulary .*vocab.ende.spm" vocab.joint.log
# Extract a textual vocabulary and compare with the expected output
LC=UTF-8 $MRT_MARIAN/spm_export_vocab --model vocab.joint/vocab.ende.spm | sort > vocab.joint.out
$MRT_TOOLS/diff-nums.py vocab.joint.out vocab.joint.expected -o vocab.joint.diff
# Exit with success code
exit 0
|
