Hi Moses-Support I firstly build the lm as below:
(base) aaronhan@grove2:~/Moses$ head -20 /home/aaronhan/Moses/lm/wmt18_5maaZhEn.arpa.en \data\ ngram 1=553697 ngram 2=8790963 ngram 3=28838697 \1-grams: -6.993945 <unk> 0 0 <s> -1.6929523 -2.0611665 </s> 0 -2.3942525 the -0.9461817 -3.6470015 show -0.6162386 -4.1808434 stars -0.5792577 -4.1823444 X -0.53191173 -5.0213814 Girls -0.33374956 -2.5064495 - -0.6326866 -2.5754578 a -0.9741243 -5.1265607 troupe -0.3527482 -2.2843323 of -0.91076595 -4.663717 talented -0.3983 -5.2061567 topless -0.27003714 Then, I binary it as bellow: /home/aaronhan/Moses/mosesdecoder/bin/build_binary /home/aaronhan/Moses/lm/wmt18_5maaZhEn.arpa.en /home/aaronhan/Moses/lm/wmt18_5maaZhEn.blm.en (base) aaronhan@grove2:~/Moses$ ls -lh /home/aaronhan/Moses/lm/wmt18_5maaZhEn.blm.en -rw-r--r-- 1 aaronhan users 715M Jul 29 18:38 /home/aaronhan/Moses/lm/wmt18_5maaZhEn.blm.en when I test it, it looks working out: (tensorflow) aaronhan@grove2:~/Moses/working$ echo "is this an English sentence ?" \ > | /home/aaronhan/Moses/mosesdecoder/bin/query /home/aaronhan/Moses/lm/wmt18_5maaZhEn.blm.en is=23 2 -2.5766807 this=321 3 -1.0483611 an=108 3 -2.1032705 English=1731 2 -2.8970845 sentence=3266 3 -2.8317716 ?=815 2 -2.0824053 </s>=2 3 -0.035550244 Total: -13.575124 OOV: 0 Perplexity including OOVs: 86.9567697150713 Perplexity excluding OOVs: 86.9567697150713 OOVs: 0 Tokens: 7 Name:query VmPeak:748300 kB VmRSS:4052 kB RSSMax:503644 kB user:0.008 sys:0.652 CPU:0.66 real:61.9797 However, when I run the overall Moses training, it produced error as below: *Wrong format of --lm. Expected: --lm factor:order:filename* (tensorflow) aaronhan@grove2:~/Moses/working$ nohup nice /home/aaronhan/Moses/mosesdecoder/scripts/training/train-model.perl -root-dir train -corpus /home/aaronhan/NMT/THUMT/corpora/preprocessed_wmt18en5maa.true -f zh -e en -alignment grow-diag-final-and -reordering msd-bidirectional-fe -lm 0:3$HOME/Moses/lm/wmt18_5maaZhEn.blm.en:8 -external-bin-dir /home/aaronhan/Moses/mosesdecoder/tools/ >& training.out & [1] 23346 GNU nano 2.7.4 File: training.out nohup: ignoring input Using SCRIPTS_ROOTDIR: /home/aaronhan/Moses/mosesdecoder/scripts Using single-thread GIZA using gzip ERROR: Wrong format of --lm. Expected: --lm factor:order:filename at /home/aaronhan/Moses/mosesdecoder/scripts/training/train-model.perl line 605. Can you let me know how to solve this, please? Kind regards, Aaron -- PhD student, Msc. (excellent), Bsc. (Maths) Google-Scholar <https://scholar.google.nl/citations?user=_vf3E2QAAAAJ&hl=en> , Presentation <https://www.slideshare.net/AaronHanLiFeng>-slides, Research-Gate <https://www.researchgate.net/profile/Aaron_L-F_Han> Google-site <https://sites.google.com/view/poetgarden/home> Linkedin <https://www.linkedin.com/in/aaronhan/>, ADAPT <https://www.adaptcentre.ie/> ResearchCentre & School of Computing, DCU, Dublin, Ireland
_______________________________________________ Moses-support mailing list Moses-support@mit.edu http://mailman.mit.edu/mailman/listinfo/moses-support