Token-level toxic span detection on URTOX (test)
60F1-ScoreXLM-RoBERTa+CRF
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| XLM-RoBERTa+CRFCRF=true2026.03 | 60 | 61 | 59 | — | |
| XLM-R+CRFLanguage=Urdu, Eval Metric=Token-level F1, Dataset=14,3422026.03 | 60 | — | — | — | |
| XLM-RoBERTaCRF=false2026.03 | 59 | 60 | 58 | 0.023 | |
| BiLSTMCRF=false2026.03 | 56 | 57 | 55 | 0.001 | |
| mBERTCRF=false2026.03 | 56 | 57 | 55 | 0.001 | |
| mBERTLanguage=Urdu, Eval Metric=Token-level F1, Dataset=14,3422026.03 | 56 | — | — | — | |
| BiLSTM-CRFLanguage=Urdu, Eval Metric=Token-level F1, Dataset=14,3422026.03 | 55.6 | — | — | — |