Named Entity Recognition on WNUT 2016 (test)
55.8F1 ScoreXLM-R-large
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| XLM-R-largeLexical Normalization=soft2020.05 | 55.8 | — | — | — | |
| RoBERTa-largeLexical Normalization=soft2020.05 | 55.4 | — | — | — | |
| XLM-R-largeLexical Normalization=hard2020.05 | 55.3 | — | — | — | |
| SANERaugmentation module=true, gate module=true2020.10 | 55.01 | — | — | — | |
| RoBERTa-largeLexical Normalization=hard2020.05 | 54.8 | — | — | — | |
| Devlin et al.2020.10 | 54.36 | — | — | — | |
| Yan et al.2020.10 | 54.06 | — | — | — | |
| Zhou et al.2020.10 | 53.43 | — | — | — | |
| HighGEN + RoSTERDictionary=Pseudo2022.10 | 53.4 | — | 49 | 58.8 | |
| DATNetLexical Normalization=soft2020.05 | 53 | — | — | — | |
| CambridgeLTLLexical Normalization=soft2020.05 | 52.4 | — | — | — | |
| BERTweetLexical Normalization=soft2020.05 | 52.1 | — | — | — | |
| GeNER + RoSTERDictionary=Pseudo2022.10 | 51.8 | — | 49.1 | 54.8 | |
| BERTweetLexical Normalization=hard2020.05 | 51.3 | — | — | — | |
| XLM-R-baseLexical Normalization=soft2020.05 | 49.9 | — | — | — | |
| RoBERTa-baseLexical Normalization=soft2020.05 | 49.7 | — | — | — | |
| XLM-R-baseLexical Normalization=hard2020.05 | 49.4 | — | — | — | |
| RoBERTa-baseLexical Normalization=hard2020.05 | 49.2 | — | — | — | |
| StandardDictionary=Full2022.10 | 45.3 | — | 44.3 | 46.2 | |
| GeNER + BONDDictionary=Pseudo2022.10 | 42.1 | — | 36.7 | 49.5 | |
| HighGEN + BONDDictionary=Pseudo2022.10 | 40.2 | — | 32.6 | 52.3 | |
| HighGEN + StandardDictionary=Pseudo2022.10 | 38.5 | — | 42.2 | 35.4 | |
| GeNER + StandardDictionary=Pseudo2022.10 | 36.5 | — | 41.3 | 32.6 | |
| HighGEN + RoSTERDictionary=Pseudo, Ablation=w/o L2022.10 | 35.1 | — | 24.3 | 63.6 | |
| BONDDictionary=Full2022.10 | 35 | — | 30.6 | 40.9 | |
| RoSTERDictionary=Full2022.10 | 28.9 | — | 43.1 | 21.8 | |
| Devlin et al.re-implemented=true2020.10 | — | 49.02 | — | — | |
| SANERAttentive semantic augmentation (AU)=true, Gate module (GA)=true2020.10 | — | 51.27 | — | — | |
| SANER (Baseline)Attentive semantic augmentation (AU)=false, Gate module (GA)=false2020.10 | — | 49.04 | — | — | |
| Yan et al.re-implemented=true2020.10 | — | 48.97 | — | — |