Stability Prediction
Collection
This group contains all relative training datasets and model checkpoints trained on these datasets for Stability Prediction.
•
2 items
•
Updated
Base model: westlake-repl/SaProt_650M_AF2
Task type: protein-level regression
Label measuring the most extreme circumstances in which protein maintains its fold above a concentration threshold (a proxy for intrinsic stability).
Dataset: SaProtHub/Dataset-Stability-TAPE
Model input type: Amino acid sequence
Performance (on test set): 0.825 Spearman's ρ
LoRA config:
- r: 8
- lora_dropout: 0.1
- 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