https://github.com/ignacioct/morphcompleteproject

Repo for the back translation project done as a COMPLETE project for Computational Morphology. We use back translation techniques to create synthetic data for training morphological inflection systems in the Danish language.
https://github.com/ignacioct/morphcompleteproject

inflection morphology multilabel-classification nlp transformers

Last synced: 27 days ago
JSON representation

Repo for the back translation project done as a COMPLETE project for Computational Morphology. We use back translation techniques to create synthetic data for training morphological inflection systems in the Danish language.

• Host: GitHub
• URL: https://github.com/ignacioct/morphcompleteproject
• Owner: ignacioct
• License: gpl-3.0
• Created: 2023-12-11T17:24:49.000Z (11 months ago)
• Default Branch: main
• Last Pushed: 2024-01-24T09:21:57.000Z (10 months ago)
• Last Synced: 2024-01-24T23:56:18.988Z (10 months ago)
• Topics: inflection, morphology, multilabel-classification, nlp, transformers
• Language: Jupyter Notebook
• Homepage:
• Size: 1.34 MB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# Backtranslation Project, by Ignacio Talavera

This is the repository for the backtranslation project done by Ignacio Talavera Cepeda in the Morphology class of 2023/2024, at HAPLAP EHU. All the training scripts I've used, as well as the training data obtained from CoNLL–SIGMORPHON 2017 can be found here. All the models were trained using Google Colaboratory default GPUs environment, therefore all the installation dependencies can be found in the individual notebooks and can be directly installed when using these environment, or can be reproduced locally if you have enough computing resources (something equivalent to Tesla K80 GPU with 12GB of RAM should be enough).