LegalGraphRAG

https://github.com/xmudeeplit/legalgraphrag
Admin
LegalGraphRAG is an evaluation framework for legal judgment prediction that integrates multi-agent...
Tokens:1,212
Snippets:15
Trust Score:9.7
Update:3 months ago
Show doc for...
Tokens:
Raw
### Install Dependencies and Configure Environment

Source: https://github.com/xmudeeplit/legalgraphrag/blob/main/README.md

Installs project dependencies and copies the example environment file. Remember to edit the `.env` file with your specific settings.

```bash
pip install -r requirements.txt

cp env.example .env
# Edit .env with model paths, API keys, and runtime settings
```

--------------------------------

### Prepare Data for CAIL Example

Source: https://github.com/xmudeeplit/legalgraphrag/blob/main/README.md

Prepares all required data for the CAIL example by running a Python script. Ensure raw data files are placed in the `./raw_data/` directory.

```bash
python scripts/prepare_data.py --dotenv-path .env --raw-data-dir ./raw_data
```

--------------------------------

### Example Output Summary Structure

Source: https://github.com/xmudeeplit/legalgraphrag/blob/main/README.md

Illustrates the structure of the output summary file generated after a successful evaluation. This includes model and dataset information, performance metrics, and output file paths.

```json
{
  "model_name": "qwen3",
  "dataset": "CAIL",
  "total_cases": 1000,
  "correct_count": 850,
  "elapsed_time": 3600.0,
  "output_file": "./outputs/CAIL/qwen3_results_combined.json"
}
```

--------------------------------

### Run Evaluation with Specified Model and Datasets

Source: https://github.com/xmudeeplit/legalgraphrag/blob/main/README.md

Executes the evaluation pipeline using a specified model and dataset. This command can be configured with various arguments to control the evaluation process.

```bash
python run.py --model qwen3 --datasets CAIL --devices cuda:2 cuda:3
```

--------------------------------

### Run with Multi-GPU

Source: https://github.com/xmudeeplit/legalgraphrag/blob/main/README.md

Execute the run script on multiple GPUs by specifying device IDs. Cases are automatically distributed.

```bash
python run.py --model qwen3 --datasets CAIL --devices cuda:0 cuda:1 cuda:2 cuda:3
```

--------------------------------

### Project Structure Overview

Source: https://github.com/xmudeeplit/legalgraphrag/blob/main/README.md

Provides a hierarchical view of the LegalGraphRAG project's directories and core modules. Useful for understanding the organization of the codebase.

```text
LegalGraphRAG/
├── core/                      # Core modules
│   ├── LegalGraphRAG.py       # Main LegalGraphRAG class
│   ├── models/                # Model implementations
│   │   ├── transformers/      # Transformers-based models (Qwen, InternLM, GLM, Gemma)
│   │   └── openai/            # OpenAI-compatible models (DeepSeek, GPT)
│   ├── graph_construct/       # Graph construction and management
│   ├── judge/                 # Legal judgment modules
│   ├── preprocess/            # Data preprocessing
│   ├── prompt/                # Prompt templates
│   └── utils/                 # Utility functions
├── scripts/                   # Data preparation scripts
├── raw_data/                  # User-provided source files for preprocessing
├── datas/                     # Generated preprocessing outputs
├── run.py                     # Main evaluation script
├── env.example                # Configuration file template
└── README.md                  # Project documentation
```
Tokens:
Raw