OLID dataset readme:
All possible label combinations across all the sub-tasks are as below: Not offensive Offensive, Untargeted Offensive, Targeted Insult, (Individual | Group | Other)
- install required packages
pip install -r requirements.txt
Baselines:
- roberta subtask_a : https://colab.research.google.com/drive/1llbEUgRT4Y6NN6_WPb6xsx207envtfhB?usp=sharing
- roberta subtask_b : https://colab.research.google.com/drive/1GoPz09K8E8RwjvWukjcz36TT7jMuH6R7?usp=sharing
- roberta subtask_c : https://colab.research.google.com/drive/1J7DlYSq17ijaoshLhNNSKdD7iZk6YUf0?usp=sharing
- Multiclass : https://colab.research.google.com/drive/1ZVEY2ddEZnzN95f7Uimq5-mJjxFfCwG8?usp=sharing [Incomplete]