Table 3 Class label classification performance on our whole MTL dataset
From: Collectively encoding protein properties enriches protein language models
Model | Pre | Acc | Rec | F1 |
|---|---|---|---|---|
BERT | 0.232 | 0.186 | 0.219 | 0.225 |
MT-BERT | 0.565 | 0.557 | 0.593 | 0.579 |
MT-BCNN | 0.716 | 0.736 | 0.667 | 0.691 |
MT-BLSTM | 0.745 | 0.726 | 0.717 | 0.731 |