LT@Helsinki at SemEval-2020 Task 12: Multilingual or language-specific BERT?

Name: LT@Helsinki at SemEval-2020 Task 12: Multilingual or language-specific BERT?
Uploaded: 2026-10-16T15:51:01.879809+00:00
Duration: 322 s
Description: The team tested different BERT models for offensive language identification and target identification across several languages for SemEval 2020 Task 12. They found that while BERT performs well generally, language-specific models like Nordic BERT achieved superior results for Danish data. Class imbalance in the target identification task also proved challenging.

The team tested different BERT models for offensive language identification and target identification across several languages for SemEval 2020 Task 12. They found that while BERT performs well generally, language-specific models like Nordic BERT achieved superior results for Danish data. Class imbalance in the target identification task also proved challenging.

Open the full episode