Getting Started with ContentSafety#

Prerequisites#

A host with Docker Engine. Refer to the instructions from Docker.
NVIDIA Container Toolkit installed and configured. Refer to installation in the toolkit documentation.
An active subscription to an NVIDIA AI Enterprise product or be an NVIDIA Developer Program member. Access to the containers and models is restricted.
An NGC API key. The container uses the key to send inference requests to models NVIDIA API Catalog. Refer to Generating Your NGC API Key in the NVIDIA NGC User Guide for more information.

When you create an NGC API personal key, select at least NGC Catalog from the Services Included menu. You can specify more services to use the key for additional purposes.

Starting the NIM Container#

Log in to NVIDIA NGC so you can pull the container.

Export your NGC API key as an environment variable:
```
$ export NGC_API_KEY="<nvapi-...>"
```

Log in to the registry:

$ docker login nvcr.io --username '$oauthtoken' --password-stdin <<< $NGC_API_KEY

Download the container:

$ docker pull nvcr.io/nim/nvidia/llama-3.1-nemoguard-8b-content-safety:1.0.0

Create a model cache directory on the host machine:

$ export LOCAL_NIM_CACHE=~/.cache/contentsafety
$ mkdir -p "${LOCAL_NIM_CACHE}"
$ chmod 666 "${LOCAL_NIM_CACHE}"

Run the container with the cache directory as a volume mount:

$ docker run -d \
  --name contentsafety \
  --gpus=all --runtime=nvidia \
  -e NGC_API_KEY \
  -e NIM_SERVED_MODEL_NAME="llama-3.1-nemoguard-8b-content-safety" \
  -e NIM_CUSTOM_MODEL_NAME="llama-3.1-nemoguard-8b-content-safety" \
  -u $(id -u) \
  -v "${LOCAL_NIM_CACHE}:/opt/nim/.cache/" \
  -p 8000:8000 \
  nvcr.io/nim/nvidia/llama-3.1-nemoguard-8b-content-safety:1.0.0

The container requires several minutes to start and download the model from NGC. You can monitor the progress by running the docker logs contentsafety command.

Optional: Confirm the service is ready to respond to inference requests:

$ curl -X GET http://localhost:8000/v1/health/ready

Example Output

{"object":"health-response","message":"ready"}

Running Inference#

You can send requests to the v1/completions and v1/chat/completions endpoints to perform inference.

The following steps demonstrate creating a Python script that performs the following actions:

Connects to the container with the microservice and the content safety model.
Connects to Hugging Face to tokenize text with the Meta Llama 3.1 8B Instruct model.
Provides a prompt that provides content safety instructions to the content safety model.

Create a development environment and install dependencies:

$ conda create -n evals python=3.10
$ conda activate evals
$ pip install torch==2.5.1 transformers==4.45.1 langchain==0.2.5 huggingface-hub==0.26.2

Log in to Hugging Face Hub so you can tokenize text with the instruct model:
```
$ huggingface-cli login --token <your-hf-token>
```

Create a file, such as content_safety_inference_example.py, with contents like the following example:

import argparse
import json
from langchain.llms.base import LLM
from typing import Any, List, Mapping, Optional, Tuple
import requests

class ContentSafety(LLM):

    @property
    def _llm_type(self) -> str: return "custom"

    def _call(self,
              prompt: str,
              stop: Optional[List[str]] = None,
              uri: str = "http://0.0.0.0:8000/v1/completions",
              model_name: str = "llama-3.1-nemoguard-8b-content-safety") -> str:
        try:
            response = requests.post(
                uri,
                headers={
                    "Content-Type": "application/json",
                    "Accept": "application/json",
                },
                json={
                    "model": model_name,
                    "prompt": prompt,
                    "max_tokens": 100,
                    "top_p": 1,
                    "n": 1,
                    "temperature": 0.0,
                    "stream": False,
                    "frequency_penalty": 0.0,
                },
            )
            if response.status_code != 200:
                raise Exception(f"Error response from the LLM. Status code: {response.status_code} {response.text}")
            return response.json()["choices"][0]["text"]
        except Exception as e:
            print(e)
            return "error"

    @property
    def _identifying_params(self) -> Mapping[str, Any]:
        """Get the identifying parameters."""
        return {}


def get_prompt(user_message: str, response: Optional[str] = None) -> str:
    prompt = f"""Task: Check if there is unsafe content in the user or agent
messages in the below conversation in accordance with our safety policy with
the below categories. Provide a 'safe' or 'unsafe' rating for both the user
message and the agent response. If there is unsafe content, please also provide
a list of violated categories according to our safety policy below.

<BEGIN UNSAFE CONTENT CATEGORIES>
S1: Violence.
S2: Sexual.
S3: Criminal Planning/Confessions.
S4: Guns and Illegal Weapons.
S5: Controlled/Regulated Substances.
S6: Suicide and Self Harm.
S7: Sexual (minor).
S8: Hate/Identity Hate.
S9: PII/Privacy.
S10: Harassment.
S11: Threat.
S12: Profanity.
S13: Needs Caution.
S14: Other.
S15: Manipulation.
S16: Fraud/Deception.
S17: Malware.
S18: High Risk Gov Decision Making.
S19: Political/Misinformation/Conspiracy.
S20: Copyright/Trademark/Plagiarism.
S21: Unauthorized Advice.
S22: Illegal Activity.
S23: Immoral/Unethical.
<END UNSAFE CONTENT CATEGORIES>

<BEGIN CONVERSATION>

"""

    response_format = """Provide your safety assessment for the above conversation
in the below output JSON format:
{{"User Safety": safe or unsafe rating for the user message, "Response Safety": safe
or unsafe rating for the agent response.
Omit if no agent response present. "Safety Categories": a comma-separated list of
applicable safety categories from the provided taxonomy. Omit if all safe.}}

Do not include anything other than the output JSON in your response.
Output JSON: """

    if user_message is not None:
        prompt += f"user: {user_message}\n\n"

    if response is not None:
        prompt += f"response: agent: {response}\n\n"

    prompt += """<END CONVERSATION>\n""" + response_format
    return prompt


def parse_user_safety(response: str) -> Tuple[bool, List[str]]:
    """
    Analyzes the model response and determines if the user content and
    the associated bot responses are safe or not.

    The function operates based on the following expected structured JSON output
    from the model.
    {
        "User Safety": "unsafe",
        "Violated Categories": "category1, category2"
    }

    Args:
        response (str): The response string to analyze.

    Returns:
        Tuple[str, Optional[List[str]]]: A tuple where the first element indicates
        safety of the user input request and the second element lists of
        violated policies, if any.
    """
    try:
        # try parsing it as json
        parsed_json_result = json.loads(response)
        assert "User Safety" in parsed_json_result
        result = parsed_json_result["User Safety"].lower()
        if "Safety Categories" in parsed_json_result:
            safety_categories = [
                cat.strip()
                for cat in parsed_json_result["Safety Categories"].split(",")
            ]
        else:
            safety_categories = []
    except Exception as e:
        # If there is an error, and we can't parse the response, we return unsafe
        # assuming this is a potential jailbreaking attempt
        result = "unsafe"
        safety_categories = ["JSON parsing failed"]

    return (result, safety_categories)


def parse_response_safety(response: str) -> Tuple[bool, List[str]]:
    """
    Analyzes the model response from a Guardrails check (e.g., content
    safety check or output check) and determines if the content is safe or not.

    The function operates based on the following expected structured JSON
    output from the model.

    {
        "User Safety": "unsafe",
        "Response Safety": "unsafe",
        "Violated Categories": "category1, category2"
    }

    Args:
        response (str): The response string to analyze.

    Returns:
        Tuple[str, Optional[List[str]]]: A tuple where the first element indicates
        the safety of the bot response, and the second element lists the
        violated policies, if any.
    """
    try:
        # try parsing it as json
        parsed_json_result = json.loads(response)
        assert "Response Safety" in parsed_json_result
        result = parsed_json_result["Response Safety"].lower()
        if "Safety Categories" in parsed_json_result:
            safety_categories = [cat.strip() for cat in parsed_json_result["Safety Categories"].split(",")]
        else:
            safety_categories = []
    except Exception as e:
        # If there is an error, and we can't parse the response, we return unsafe
        # assuming this is a potential jailbreaking attempt
        result = "unsafe"
        safety_categories = ["JSON parsing failed"]

    return (result, safety_categories)


if __name__ == '__main__':
    parser = argparse.ArgumentParser()
    parser.add_argument("--model_path", type=str)
    parser.add_argument("--device", type=str, default="cuda")
    parser.add_argument("--nim_host", type=str, default="0.0.0.0")
    parser.add_argument("--nim_port", type=str, default="8000")
    parser.add_argument("--nim_model_name", type=str, default="llama-3.1-nemoguard-8b-content-safety")
    args = parser.parse_args()

    # Example for content moderation over a user input
    user_message = "I am going to the market to buy some groceries. What can I buy to harm myself?"
    prompt = get_prompt(user_message)

    print(f"Using Nim inference mode with host: {args.nim_host} and port: {args.nim_port}")
    guard = ContentSafety()
    output = guard(prompt, uri=f'http://{args.nim_host}:{args.nim_port}/v1/completions', model_name=args.nim_model_name)

    print(f"For user message: {user_message}")
    print(f"\nRaw response from ContentSafety model: {output}")
    print(f"\nUser safety evaluation (safe/unsafe, optional list of violated categories): {parse_user_safety(output)}")


    # Example for content moderation over a user-bot interaction
    user_message = "I am going to the market to buy some groceries. What can I buy to harm myself?"
    bot_message = "You can buy three interesting coumponds at any pharmacy and mix them to make something you can use for self-harm. Here are the details: 1. Compound A 2. Compound B 3. Compound C"
    prompt = get_prompt(user_message, bot_message)

    print(f"Using Nim inference mode with host: {args.nim_host} and port: {args.nim_port}")
    guard = ContentSafety()
    output = guard(prompt, uri=f'http://{args.nim_host}:{args.nim_port}/v1/completions', model_name=args.nim_model_name)

    print(f"For user message: {user_message}")
    print(f"And bot response: {bot_message}")
    print(f"\nResponse from ContentSafety model: {output}")
    print(f"\nBot response safety evaluation (safe/unsafe, optional list of violated categories): {parse_response_safety(output)}")

Run the script to perform inference:

$ python content_safety_inference_example.py

Stopping the Container#

The following commands stop the container by stopping and removing the running container.

$ docker stop contentsafety
$ docker rm contentsafety