wubet/amharic-english-machine-translation-transformer-baseline

# Amharic-English-Machine-Translation-Transformer-Baseline clone the " amharic-english-machine-translation-baseline" repository. using SSH command. ```buildoutcfg git clone git@github.com:wubet/amharic-english-machine-translation-transformer-baseline.git ``` Or using the https command ```buildoutcfg git clone github.com ``` Change the directory into the main repository. ```buildoutcfg cd amharic-english-machine-translation-transformer-baseline ``` Clone the "unified-amharic-english-corpus" repository that contains Amharic-English dataset. using SSH command. ```buildoutcfg git clone github.com ``` Clone the "tf-transformer " repository that is a TensorFlow 2.x implementation of Transformer model (Attention is all you need) for Neural Machine Translation (NMT) authored by Chao ji. using SSH command. ```buildoutcfg git clone git@github.com:chao-ji/tf-transformer.git ``` Or using the https command ```buildoutcfg git clone github.com ``` Change directory to Update the submodule ```buildoutcfg cd tf-transformer ``` Update the submodule by running ```buildoutcfg git submodule update --init --recursive ``` Return to main directory ```buildoutcfg cd .. ``` Execute the pre-process file ```buildoutcfg python pre-process.py ``` We need to transliterate the Gee'z character representation into latin character representation. For development or validation ```buildoutcfg python3 tf-transformer/commons/create_transliteration.py \ --original_filenames=tf-transformer/unified-amharic-english-corpus/datasets/dev.am-en.base.am \ --transliterate_filenames=tf-transformer/unified-amharic-english-corpus/datasets/dev.am-en.transliteration.am \ ``` For training ```buildoutcfg python3 tf-transformer/commons/create_transliteration.py \ --original_filenames=tf-transformer/unified-amharic-english-corpus/datasets/train.am-en.base.am …