Skip to content

Latest commit

 

History

History
15 lines (10 loc) · 788 Bytes

File metadata and controls

15 lines (10 loc) · 788 Bytes

Stylized-Neural-Machine-Translation

In this project, we are interested in improving the readability of archaic Shakespearean texts by implementing neural machine translation of text followed by style transfer to convert the input into formally correct modern English. We have two constituent models - a Neural Machine Translator and a text style transfer model, and perform multiple experiments to improve the readability metric of the input text.

In this project we have trained 5 models

  1. Baseline
  2. translator
  3. pre-trained-formality-transfer-main
  4. augmented-data-translator
  5. StylizedNMT

Individual README is under each folder.

Complete translation dataset present in : https://docs.google.com/spreadsheets/u/1/d/1F_wnhke4Ebaa_fz0q3cRwDCPcSS8sDCBjB-tEM4tyvU/edit#gid=0