Dataset description: SemEval-2018 task 3 - Irony detection in English tweets
-
Project report is in folder /doc
-
Install requirements first
-
Run models/task3/train for ntua implementation
-
Run /Vandad/try-bert to play with bert
-
Download ntua embeddings from their github repository by running one of their tasks and put it under "embeddings"
-
Results are in foler /out/experiments/
The follwing is our proposed architecture:
