chasemcdougall/capstone-api repository overview

⁠OpenAI Clip Adapter

This repository stores the Docker images associated with an api for wrapping a caching adapter method for OpenAI's Clip. The implementation code can be found here⁠

This image is dependent on a RedisAI instance also running and accessible. If you wish to swap this out for some other form of Redis you can edit the connection interface⁠.

⁠To boot the image as is you can run the following:

1. Start the RedisAI container:

docker run -p 6379:6379 -d redislabs/redisai

1. Start this container:

docker run --gpus all -p 80:80 chasemcdougall/capstone-api

1. You can now interact with the service through its API with interactive documentation available at http://localhost:80/docs⁠

⁠Considerations

• This image is dependent on the host device having cuda-enabled graphics.
• Assumes containers have access to host.docker.internal, if you are using an alternative docker provider such as Lima you'll likely want to setup a network for communication and pass in a new REDIS_URL via -e 'REDIS_URL=redis://redis-0:6379'

⁠Setting up a Network (Lima and nerdctl implementations)

1. Create a docker network

docker network create capstone_network

1. Add your Redis container to the network

docker run -p 6379:6379 -d --network capstone_network --name redis-0 redislabs/redisai

1. Add your main container to the network through one of the following methods and pass in the Redis

docker run --gpus all -p 80:80 --network capstone_network -e "REDIS_URL=redis://redis-0:6379" chasemcdougall/capstone-api:1.1.0