diff options
Diffstat (limited to 'gi/pipeline/lticluster.config')
| -rw-r--r-- | gi/pipeline/lticluster.config | 9 |
1 files changed, 9 insertions, 0 deletions
diff --git a/gi/pipeline/lticluster.config b/gi/pipeline/lticluster.config new file mode 100644 index 00000000..3e23c8cb --- /dev/null +++ b/gi/pipeline/lticluster.config @@ -0,0 +1,9 @@ +# THIS FILE GIVES THE LOCATIONS OF THE CORPORA USED +# name path aligned-corpus LM dev dev-refs test1 testt-eval.sh ... +/home/cdyer/ws10smt-data +btec /home/cdyer/ws10smt-data/btec/ split.zh-en.al lm/en.3gram.lm.gz devtest/devset1_2.zh devtest/devset1_2.lc.en* devtest/devset3.zh eval-devset3.sh +zhen /home/cdyer/ws10smt-data/chinese-english corpus.zh-en.al lm/c2e.3gram.lm.gz dev_and_test/mt02.src.txt dev_and_test/mt02.ref.* dev_and_test/mt03.src.txt eval-mt03.sh +aren /home/cdyer/ws10smt-data/arabic-english corpus.ar-en-al lm/a2e.3gram.lm.gz dev_and_test/dev.src.txt dev_and_test/dev.ref.txt.* dev_and_test/mt05.src.txt eval-mt05.sh +uren /home/cdyer/ws10smt-data/urdu-english corpus.ur-en.al lm/u2e.en.lm.gz dev/dev.ur dev/dev.en* devtest/devtest.ur eval-devtest.sh +nlfr /home/cdyer/ws10smt-data/dutch-french corpus.nl-fr.al + |