Table 6 Comparison of TAPE benchmark results based solely on family and superfamily classification task
From: Collectively encoding protein properties enriches protein language models
Model | Structure | Evolution | |
|---|---|---|---|
SS prediction | Contact prediction | Remote homology | |
MT-BERT | 0.71 | 0.34 | 0.21 |
MT-BCNN | 0.75 | 0.35 | 0.23 |
MT-BLSTM | 0.72 | 0.38 | 0.25 |