Table 7 Eight scenarios for selective fine-tuning
From: AMPDeep: hemolytic activity prediction of antimicrobial peptides using transfer learning
Scenario | Description |
|---|---|
1 | Full fine-tuning |
2 | Feature extraction |
3 | Pooler fine-tuning |
4 | Pooler replacement |
5 | Input embedding fine-tuning |
6 | Positional embedding fine-tuning |
7 | Input embedding + layer norm fine-tuning |
8 | Positional embedding + layer norm fine-tuning |