Skip to content

hardikyagnik/CharNMT

Repository files navigation

Character Level Neural Machine Translation

Done as a part of an academic project for CS 583 - Deep Learning

Task: To train a model to transform English words to their transformed forms as per the rules bellow

Transformation Rules

  • If the first letter is a consonant, then that letter is moved to the end of the word, and “ay” is appended, e.g., slow → lowsay
  • If a word starts with a vowel, then append “way” at the end, e.g., amoeba → amoebaway
  • Some consonant pairs like “sh” are moved together to the end of the word with “ay” appended, e.g., shallow → allowshay

You can train the model by executing the python file execute_models.py and, you can see the results in the output/folder. After each epoch of model training, you can find the translation of the sentence "the air conditioning is working".

About

Character level Neural Machine Translation

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published