Review 2023-05-02 + VM run

[igornishka]

• you can review readme before pushing e.g. here
• dump intermediate data e.g. the 20M sampled sentences
• fix passing arguments via the script to be able to better control the file naming

[igornishka]

During the clean installation and running on the VM, I encountered the following:

• wikisimple_folder = $data_folder/Model_English_C_English_S doesn't work due to the space -> wiki simple says: Saving to: '/WikiSimple.tar.gz -> tar: NMT-Data/Model_English_C_English_S/WikiSimple.tar.gz: Cannot open: No such file or directory
• sudo apt-get install unzip as it's a bit annoying, it gets flooded by the rest of the script output
• let's check if we can exit on error in this whole script or deal with the output differently
• FileNotFoundError: [Errno 2] No such file or directory: 'NMT-Data/Eval_Medical_Dutch_C_Dutch_S/NL_test_org' -> I completely forgot this one, so maybe we can make things fool-proof and check if the eval data (or anything else that's needed) is there in the beginning and exit if that's not the case?

[igornishka]

• fix no-gpu usage (https://forum.opennmt.net/t/train-model-with-cpu-only-using-python/4157)
• onmt_train -config config/model_1.yaml -src_vocab NMT-Data/Model_Dutch_C_English_C/run/.vocab.src -tgt_vocab NMT-Data/Model_Dutch_C_English_C/run/.vocab.tgt -> no dot in the vocab file names anymore