Metric: macro avg (subtask 1) (higher is better)
| # | Model↕ | macro avg (subtask 1)▼ | Extra Data | Paper | Date↕ | Code |
|---|---|---|---|---|---|---|
| 1 | Subword-ULM transformer (DeepSPIN-3; soft-attention, 1-5 entmax) | 97.29 | No | - | - | - |
| 2 | Char LSTM (DeepSPIN-2; soft-attention, 1-5 entmax) | 97.15 | No | - | - | - |
| 3 | Ensemble of hard-attention transducers (CLUZH) | 96.85 | No | - | - | Code |
| 4 | Char LSTM (DeepSPIN-1; soft-attention) | 96.32 | No | - | - | - |
| 5 | BiLSTM for seq labelling (Tü_Seg-1) | 96.06 | No | - | - | - |
| 6 | Bidirectional GRU + Morfessor features (AUUH_F) | 93.72 | No | - | - | - |
| 7 | HMM (JB132; trained w EM) | 58.39 | No | - | - | - |
| 8 | Morfessor2 | 25.57 | No | The SIGMORPHON 2022 Shared Task on Morpheme Segm... | 2022-06-15 | Code |
| 9 | ULM | 20.61 | No | The SIGMORPHON 2022 Shared Task on Morpheme Segm... | 2022-06-15 | Code |
| 10 | WordPiece | 15.89 | No | The SIGMORPHON 2022 Shared Task on Morpheme Segm... | 2022-06-15 | Code |