|
|
|
|
|
--- |
|
|
|
|
|
base_model: ProTrekHub/Protein_Encoder_35M |
|
|
|
|
|
library_name: peft |
|
|
|
|
|
--- |
|
|
|
|
|
|
|
|
|
|
|
# Model Card for Model-RASH-ProTrek-35M |
|
|
<slot name='description'> |
|
|
|
|
|
## Task type |
|
|
Protein-level Regression |
|
|
|
|
|
**Dataset:** [SeprotHub/Dataset-RASH_HUMAN](https://huggingface.co/datasets/SeprotHub/Dataset-RASH_HUMAN) |
|
|
## Model input type |
|
|
AA Sequence |
|
|
|
|
|
## LoRA config |
|
|
|
|
|
- **r:** 8 |
|
|
- **lora_dropout:** 0.0 |
|
|
- **lora_alpha:** 16 |
|
|
- **target_modules:** ['key', 'output.dense', 'value', 'intermediate.dense', 'query'] |
|
|
- **modules_to_save:** ['classifier'] |
|
|
|
|
|
## Training config |
|
|
|
|
|
- **optimizer:** |
|
|
- **class:** AdamW |
|
|
- **betas:** (0.9, 0.98) |
|
|
- **weight_decay:** 0.01 |
|
|
- **learning rate:** 0.0005 |
|
|
- **epoch:** 1 |
|
|
- **batch size:** 32 |
|
|
- **precision:** 16-mixed |
|
|
|
|
|
|
