Instructions to use narcolepticchicken/safelawbench-model with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use narcolepticchicken/safelawbench-model with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="narcolepticchicken/safelawbench-model")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("narcolepticchicken/safelawbench-model")
model = AutoModelForCausalLM.from_pretrained("narcolepticchicken/safelawbench-model")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use narcolepticchicken/safelawbench-model with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "narcolepticchicken/safelawbench-model"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "narcolepticchicken/safelawbench-model",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/narcolepticchicken/safelawbench-model

SGLang

How to use narcolepticchicken/safelawbench-model with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "narcolepticchicken/safelawbench-model" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "narcolepticchicken/safelawbench-model",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "narcolepticchicken/safelawbench-model" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "narcolepticchicken/safelawbench-model",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use narcolepticchicken/safelawbench-model with Docker Model Runner:
```
docker model run hf.co/narcolepticchicken/safelawbench-model
```

Model Card for SafeLawBench-Specialized Model

This model is a fine-tuned version of Qwen/Qwen3-8B, specialized for legal safety tasks on the SafeLawBench benchmark.

It was trained on synthetic legal safety multiple-choice questions covering Mainland China and Hong Kong SAR legal frameworks, including civil law, criminal law, privacy law, domestic violence law, employment law, consumer protection, and family/child law.

WARNING: This is a partial checkpoint (500 training steps). It is NOT the final model. A fully trained version is in development and will replace this artifact.

Training Data

Trained on narcolepticchicken/safelawbench-synthetic, a synthetic dataset of legal safety MCQs generated from public Chinese and Hong Kong legal materials. The dataset is balanced across risk categories as defined in the SafeLawBench taxonomy.

Intended Use

Legal safety evaluation on the SafeLawBench benchmark
Research on legal AI safety
NOT intended for: providing actual legal advice, production legal systems, or any use where incorrect legal guidance could cause harm

Limitations

This is a 500-step partial checkpoint — performance is lower than the final model
Trained on synthetic data, not real legal exam questions
May not generalize beyond the SafeLawBench-style multiple choice format
Has not been evaluated on real legal cases or open-ended legal questions

Quick start

from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "narcolepticchicken/safelawbench-model"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype="auto", device_map="auto")

# SafeLawBench format
messages = [
    {"role": "system", "content": "You are a legal expert. Answer with [[ ANSWER ]] LETTER only."},
    {"role": "user", "content": "Question text here with options A) B) C) D)"}
]
inputs = tokenizer.apply_chat_template(messages, tokenize=True, add_generation_prompt=True, return_tensors="pt").to(model.device)
outputs = model.generate(inputs, max_new_tokens=64, do_sample=False)
response = tokenizer.decode(outputs[0][len(inputs[0]):], skip_special_tokens=True)
print(response)

Training procedure

This model was trained with SFT using TRL's SFTTrainer with LoRA adapters (r=16, alpha=32).

Framework versions

TRL: 1.3.0
Transformers: 5.8.0
PyTorch: 2.11.0
Datasets: 4.8.5
Tokenizers: 0.22.2

License

Apache 2.0

Citation

@software{vonwerra2020trl,
  title   = {{TRL: Transformers Reinforcement Learning}},
  author  = {von Werra, Leandro and Belkada, Younes and Tunstall, Lewis and Beeching, Edward and Thrush, Tristan and Lambert, Nathan and Huang, Shengyi and Rasul, Kashif and Gallouédec, Quentin},
  license = {Apache-2.0},
  url     = {https://github.com/huggingface/trl},
  year    = {2020}
}

Generated by ML Intern

This model repository was generated by ML Intern, an agent for machine learning research and development on the Hugging Face Hub.

Generated by ML Intern

This model repository was generated by ML Intern, an agent for machine learning research and development on the Hugging Face Hub.

Try ML Intern: https://smolagents-ml-intern.hf.space
Source code: https://github.com/huggingface/ml-intern

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = 'narcolepticchicken/safelawbench-model'
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id)

For non-causal architectures, replace AutoModelForCausalLM with the appropriate AutoModel class.

Downloads last month: 57

Safetensors

Model size

8B params

Tensor type

BF16

Model tree for narcolepticchicken/safelawbench-model

Base model

Qwen/Qwen3-8B-Base

Finetuned

Qwen/Qwen3-8B

Finetuned

(1572)

this model