Argo API Bridge

This project provides a compatibility layer that transforms OpenAI-style API requests into Argonne National Lab's Argo API format. It supports chat completions, text completions, and embeddings endpoints.

Downstream Integration

Several tools have been tested with the bridge, including IDE integrations, web UI's, and python libraries. Setup guides for these tools tools are located in the downstream_config.md.

Setup

1. Create Conda Environment

First, create a new conda environment with Python 3.12:

conda create -n argo_bridge python=3.12
conda activate argo_bridge

2. Install Requirements

Install the required Python packages using pip:

pip install -r requirements.txt

3. Run the Server

Start the server with default settings (port 7285):

python argo_bridge.py

Configuration Options

The server supports the following command-line arguments:

• --username: Set the username for API requests (default: 'APS')
• --port: Set the port number for the server (default: 7285)
• --dlog: Enable debug-level logging (when set, logging is at DEBUG level; by default it is INFO level)

Example with custom settings:

python argo_bridge.py --username custom_user --port 8000 --dlog

Endpoints

The API exposes the following endpoints:

• Chat Completions: /chat/completions (POST)
• Text (Legacy) Completions: /completions (POST)
• Embeddings: /embeddings (POST)

Supported Models

The server accepts both Argo and OpenAI model identifiers.

Chat and Completion Models

• GPT-3.5: (gpt35, gpt-3.5)
• GPT-3.5 Large: (gpt35large)
• GPT-4: (gpt4, gpt-4)
• GPT-4 Large: (gpt4large)
• GPT-4 Turbo: (gpt4turbo, gpt-4-turbo)
• GPT-4o: (gpt4o, gpt-4o, gpt-4o-mini)
• GPT-o1 Preview: (gpto1preview, o1-preview)
• GPT-o1 Mini: (gpto1mini, o1-mini, o1mini)
• GPT-o3 Mini: (gpto3mini, o3-mini, o3mini)

Embedding Models

• v3small: (text-embedding-3-small, v3small)
• v3large: (text-embedding-3-large, v3large)
• ada002: (text-embedding-ada-002, ada002)

Production Deployment

For personal use, the development server should be plenty, but if you wish to scale up, this project includes a docker-compose.yaml file to manage the following services:

• argo_bridge: The main application container, built using the provided dockerfile. It runs the Argo API bridge using Gunicorn.
• prometheus: A Prometheus container for collecting metrics from the argo_bridge service.
• grafana: A Grafana container for visualizing the metrics collected by Prometheus. It comes pre-configured with a dashboard for the Argo Bridge.

Steps to run

• To run, first set the environment variable METRICS_TOKEN to an arbitrary string.
• Then, copy the prometheus.yml.template to prometheus.yml replacing the bearer token to that string.
• Currently the prod setup is configured for SSL and requires a myserver.crt and myserver.key. Either generate these, or change the gunicorn and prometheus service to http.
• Once that is in place, build and run the containers using the following command from the root of the project directory:

docker-compose up -d

This will start all services in detached mode.

• The Argo Bridge will be accessible on port 80 of the host machine.
• Prometheus will be accessible on http://localhost:9090.
• Grafana will be accessible on http://localhost:3000. The default credentials are admin / admin_password.

To stop the containers:

docker-compose down

Manual Gunicorn Deployment (without Docker)

If you prefer to run the application with Gunicorn directly without Docker, you can use the following command:

gunicorn --workers 4 --bind localhost:7285 argo_bridge:app

Testing

Run the test suite using unittest:

python -m unittest test_server.py