|
--- |
|
license: mit |
|
--- |
|
|
|
**Base model:** [westlake-repl/SaProt_650M_AF2](https://huggingface.co/westlake-repl/SaProt_650M_AF2) |
|
|
|
**Task type:** protein-level regression |
|
|
|
Label is the fitness score measuring the effect of mutations on AAV (Adeno-associated virus). |
|
|
|
**Dataset:** [SaProtHub/Dataset-AAV-FLIP](https://huggingface.co/datasets/SaProtHub/Dataset-AAV-FLIP) |
|
|
|
**Model input type:** Amino acid sequence |
|
|
|
**Performance (on test set):** 0.809 Spearman's ρ |
|
|
|
**LoRA config:** |
|
- **r:** 8 |
|
- **lora_dropout:** 0 |
|
- **lora_alpha:** 8 |
|
- **target_modules:** ["query", "key", "value", "intermediate.dense", "output.dense"] |
|
- **modules_to_save:** ["classifier"] |
|
|
|
**Training config:** |
|
|
|
- **optimizer:** |
|
- **class:** AdamW |
|
- **betas:** (0.9, 0.98) |
|
- **weight_decay:** 0.01 |
|
- **learning rate:** 1e-4 |
|
- **epoch:** 50 |
|
- **batch size:** 64 |
|
- **precision:** 16-mixed |
