Text Classification on Yahoo! Answer 2000 labels
29.11Top-1 Error Rate (%)SoftMatch/FreeMatch + SemiReward
Evaluation results
| Method | Links | ||
|---|---|---|---|
| SoftMatch/FreeMatch + SemiRewardBackbone=Bert [Devlin et al., 2018], SemiReward=true2023.10 | 29.11 | 0.99 | |
| FlexMatch + SemiRewardBackbone=Bert [Devlin et al., 2018], SemiReward=true2023.10 | 29.97 | 0.99 | |
| SimMatch# Labels=2000, Backbone=Bert-Base2023.05 | 30.16 | — | |
| CoMatch# Labels=2000, Backbone=Bert-Base2023.05 | 30.25 | — | |
| FreeMatch# Labels=2000, Backbone=Bert-Base2023.05 | 30.32 | — | |
| SoftMatch/FreeMatchBackbone=Bert [Devlin et al., 2018], SemiReward=false2023.10 | 30.34 | — | |
| AdaMatch# Labels=2000, Backbone=Bert-Base2023.05 | 30.44 | — | |
| SoftMatch# Labels=2000, Backbone=Bert-Base2023.05 | 30.44 | — | |
| FixMatch# Labels=2000, Backbone=Bert-Base2023.05 | 30.51 | — | |
| ILL# Labels=2000, Backbone=Bert-Base2023.05 | 30.86 | — | |
| FlexMatchBackbone=Bert [Devlin et al., 2018], SemiReward=false2023.10 | 31.06 | — | |
| FlexMatch# Labels=2000, Backbone=Bert-Base2023.05 | 31.13 | — | |
| Dash# Labels=2000, Backbone=Bert-Base2023.05 | 31.19 | — | |
| VAT# Labels=2000, Backbone=Bert-Base2023.05 | 31.5 | — | |
| Pseudo Label + SemiRewardBackbone=Bert [Devlin et al., 2018], SemiReward=true2023.10 | 32.5 | 0.99 | |
| Pseudo-Label# Labels=2000, Backbone=Bert-Base2023.05 | 32.72 | — | |
| Pseudo LabelBackbone=Bert [Devlin et al., 2018], SemiReward=false2023.10 | 33.14 | — | |
| Mean-Teacher# Labels=2000, Backbone=Bert-Base2023.05 | 33.43 | — | |
| MixMatch# Labels=2000, Backbone=Bert-Base2023.05 | 33.62 | — |