I want to train an smt model using mosesdecoder (based on the moses baseline:https://www2.statmt.org/moses/?n=Moses.Baseline). I'm encountering the following error:
this is the command:
arisstamos@DESKTOP-DAJTUSV:~/mosesdecoder$ perl ~/mosesdecoder/scripts/training/train-model.perl \
--root-dir work.en-de
--model-dir work.en-de/model
--corpus /mnt/c/home/arisstamos/mosesdecoder/Europarl_filtered_final.txt
--f en --e el
--external-bin-dir ~/mosesdecoder/scripts/opt/GIZA++
--mgiza -mgiza-cpus 4
--parallel
--first-step 1 --last-step 3
This is the output I'm getting:
Using SCRIPTS_ROOTDIR: /home/arisstamos/mosesdecoder/scripts Using multi-thread GIZA using gzip (1) preparing corpus @ Sat Nov 18 17:05:13 EET 2023 Executing: mkdir -p /home/arisstamos/mosesdecoder/work.en-de/corpus (1.0) selecting factors @ Sat Nov 18 17:05:13 EET 2023 Forking... (1.1) running mkcls @ Sat Nov 18 17:05:13 EET 2023 /home/arisstamos/mosesdecoder/scripts/opt/GIZA++/mkcls -c50 -n2 -p/mnt/c/home/arisstamos/mosesdecoder/Europarl_filtered_final.txt.en -V/home/arisstamos/mosesdecoder/work.en-de/corpus/en.vcb.classes opt /home/arisstamos/mosesdecoder/work.en-de/corpus/en.vcb.classes already in place, reusing (1.2) creating vcb file /home/arisstamos/mosesdecoder/work.en-de/corpus/en.vcb @ Sat Nov 18 17:05:13 EET 2023 ERROR: Can't read /mnt/c/home/arisstamos/mosesdecoder/Europarl_filtered_final.txt.en at /home/arisstamos/mosesdecoder/scripts/training/train-model.perl line 975. (1.1) running mkcls @ Sat Nov 18 17:05:13 EET 2023 /home/arisstamos/mosesdecoder/scripts/opt/GIZA++/mkcls -c50 -n2 -p/mnt/c/home/arisstamos/mosesdecoder/Europarl_filtered_final.txt.el -V/home/arisstamos/mosesdecoder/work.en-de/corpus/el.vcb.classes opt /home/arisstamos/mosesdecoder/work.en-de/corpus/el.vcb.classes already in place, reusing
The directories are fine, all the files exist and I've checked the paths too. I've tried to change the file type from CSV to txt but the error persist.
What do you think that the problem might be? Thank you in advance