Papers › MIPT-NSU-UTMN at SemEval-2021 Task 5: Ensembling Learning with Pre-trained Language...

MIPT-NSU-UTMN at SemEval-2021 Task 5: Ensembling Learning with Pre-trained Language Models for Toxic Spans Detection

10 Apr 2021SEMEVAL 2021arXiv:2104.04739archive 2025-07-28

Mikhail Kotyushev, Anna Glazkova, Dmitry Morozov

This paper describes our system for SemEval-2021 Task 5 on Toxic Spans Detection. We developed ensemble models using BERT-based neural architectures and post-processing to combine tokens into spans. We evaluated several pre-trained language models using various ensemble techniques for toxic span identification and achieved sizable improvements over our baseline fine-tuned BERT models. Finally, our system obtained a F1-score of 67.55% on test data.

PaperPDFConference PDFCode

Code

morozowdmitry/semeval21 officialmentioned in paper report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Toxic Spans Detection

Results from the paper archive 2025-07-28

No leaderboard rows for this paper in the archive.

Methods

AdamAttentionAttention DropoutBERTDense ConnectionsDropoutLayer NormalizationLinear LayerLinear Warmup With Linear DecayMulti-Head AttentionResidual ConnectionSoftmaxWeight DecayWordPiece

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections