Upload README.md with huggingface_hub
Browse files
README.md
ADDED
|
@@ -0,0 +1,30 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
language:
|
| 3 |
+
- en
|
| 4 |
+
tags:
|
| 5 |
+
- legal
|
| 6 |
+
- llama
|
| 7 |
+
- lora
|
| 8 |
+
- adapter
|
| 9 |
+
- legal-assistant
|
| 10 |
+
base_model: unsloth/Meta-Llama-3.1-8B-bnb-4bit
|
| 11 |
+
library_name: peft
|
| 12 |
+
license: llama3.1
|
| 13 |
+
---
|
| 14 |
+
|
| 15 |
+
# Legal Specialist Llama 3.1 8B (LoRA Adapter)
|
| 16 |
+
|
| 17 |
+
This is the **LoRA adapter only** for legal domain specialization - requires base model.
|
| 18 |
+
|
| 19 |
+
## Model Details
|
| 20 |
+
- **Model Type**: LoRA adapter weights only
|
| 21 |
+
- **Size**: ~335MB (adapter weights only)
|
| 22 |
+
- **Training**: Fine-tuned on 41,133 legal documents
|
| 23 |
+
|
| 24 |
+
## LoRA Configuration
|
| 25 |
+
- **Rank**: 32
|
| 26 |
+
- **Alpha**: 64
|
| 27 |
+
- **Target Modules**: q_proj, k_proj, v_proj, o_proj, gate_proj, up_proj, down_proj
|
| 28 |
+
- **Dropout**: 0.0
|
| 29 |
+
|
| 30 |
+
**Note**: This is the LoRA adapter only. For the full merged model, see: [cdomond45/GTM-legal-specialist-3.1-merged](cdomond45/GTM-legal-specialist-3.1-merged)
|