Skip to content

Latest commit

 

History

History
84 lines (55 loc) · 2.66 KB

README.md

File metadata and controls

84 lines (55 loc) · 2.66 KB

eidos: Validation and execution of AI functions

Linter: Ruff Python Version from PEP 621 TOML GitHub Release GitHub Actions Workflow Status

eidos is an API for validating and executing AI functions. It aims to be a generic API to serve as a common interface to allow execution of functions by LLMs.

Usage diagram of eidos

Install

From source:

git clone [email protected]:KhaosResearch/eidos.git
cd eidos
python -m pip install -e .

Or directly from GitHub:

python -m pip install "eidos @ git+ssh://[email protected]/KhaosResearch/eidos.git"

Deployment

  • Development

Run the API with the following command:

uvicorn eidos.api:app --host 0.0.0.0 --port 8090 --reload

You can override the default configuration by setting environment variables.

  • Docker

Alternatively, you can use the provided Dockerfile to build a Docker image and run the API in a container:

docker build -t eidos-server:latest .
docker run --rm  -v $(pwd)/functions:/functions -p 8090:80 eidos-server:latest

Example:

curl -X POST -H "Content-Type: application/json" -d '{"who": "me"}' http://localhost:8090/api/v1/execution/salute
  • Kubernetes

To deploy the container in Kubernetes, a reference deployment is available and documented at manifests.

  • Serverless in AWS Another docker image to deploy serverless in AWS Lambda is provided in Dockerfile.lambda. The image is based on the official AWS Lambda Python 3.11 image. For extending this image the process is the same as the main image.
docker build -t eidos-lambda -f Dockerfile.lambda .

Run the container locally with the following command or deploy in AWS Lambda as a docker container image:

docker run --rm -p 8091:8080 eidos-lambda

Invoke the function for local testing with sample query

curl -XPOST "http://localhost:8091/2015-03-31/functions/function/invocations" -d '{"command": "EXECUTE", "parameters": {"function": "salute", "args": {"who": "me, I am executing serverless"}}}'

Testing

pytest is used for testing. You can run the tests with the following command:

pytest tests/