Views
No views yet
| Parameter | Value |
|---|---|
| Training Method | SFT |
| LoRA Rank | 32 |
| Epochs | 2 |
| Learning Rate | 3e-5 |
| Scheduler | Cosine |
| Warmup Ratio | 0.05 |
| Weight Decay | 0.02 |
| Metric | Base Model | Fine-tuned Model |
|---|---|---|
| Dataset Win Rate | 37% | 63% |
| Legal Win Rate | 31% | 69% |
Note: This model is intended as an educational and productivity tool. It should not replace professional legal advice.