mirror of
https://github.com/moses-smt/mosesdecoder.git
synced 2024-09-11 11:25:40 +03:00
When reading hypothesis and reference files, multi-bleu.perl uses the chop function to remove the trailing newline character. If one of these files happens to not end with a newline, then chop will remove the last character of the last line (instead of the newline). This causes the BLEU score to be slightly off from its theoretical value. Using the safest chomp function solves this problem, i.e. it only removes newlines when present. |
||
---|---|---|
.. | ||
analysis | ||
docker | ||
ems | ||
fuzzy-match | ||
generic | ||
nbest-rescore | ||
OSM | ||
other | ||
recaser | ||
regression-testing | ||
server | ||
share/nonbreaking_prefixes | ||
tests | ||
tokenizer | ||
training | ||
Transliteration | ||
Jamfile | ||
README |
2006-07-29 This directory should contain all multi-purpose scripts for: - training ... training moses (including BLEU evaluation needed for MERT) - analysis ... analyzing MT output (for human analysis) - generic ... script for handling generic issues (parallelization) - lib ... perl modules used by various scripts The Jamfile then takes care of proper 'release' from your git directory to the shared directories. The released scripts should remain in the *same directory structure*.