POS tagging on POS tagging (myv) Few-Text (test)
74.3F1 ScoreGold MLM + Both (Label Distillation)
Evaluation results
| Method | Links | |
|---|---|---|
| Gold MLM + Both (Label Distillation)Backbone=XLM-R base, Adaptation Method=Pseudo Trans-train + Pseudo MLM + Label Distillation, Data Setting=Few-Text2022.03 | 74.3 | |
| Gold MLM + Pseudo MLMBackbone=XLM-R base, Adaptation Method=Pseudo MLM, Data Setting=Few-Text2022.03 | 73.8 | |
| Gold MLM (Ours)Backbone=XLM-R base, Data Setting=Few-Text2022.03 | 70.6 | |
| Ebrahimi and Kann (2021)Backbone=XLM-R large, Data Setting=Few-Text2022.03 | 66.6 | |
| Gold MLM + Both (Pseudo Trans-train & MLM)Backbone=XLM-R base, Adaptation Method=Pseudo Trans-train + Pseudo MLM, Data Setting=Few-Text2022.03 | 65.9 | |
| Gold MLM + Pseudo Trans-trainBackbone=XLM-R base, Adaptation Method=Pseudo Trans-train, Data Setting=Few-Text2022.03 | 63.8 |