Harshul Surana


2022

pdf
giniUs @LT-EDI-ACL2022: Aasha: Transformers based Hope-EDI
Harshul Surana | Basavraj Chinagundi
Proceedings of the Second Workshop on Language Technology for Equality, Diversity and Inclusion

This paper describes team giniUs’ submission to the Hope Speech Detection for Equality, Diversity and Inclusion Shared Task organised by LT-EDI ACL 2022. We have fine-tuned the Roberta-large pre-trained model and extracted the last four decoder layers to build a classifier. Our best result on the leaderboard achieve a weighted F1 score of 0.86 and a Macro F1 score of 0.51 for English. We have secured a rank of 4 for the English task. We have open-sourced our code implementations on GitHub to facilitate easy reproducibility by the scientific community.