The easiest way to get started with Articulate is to use our included docker-compose file. Articulate relies on several services besides it's own UI and API. These all get started when you use the provided compose.
A few warnings before we get started:
- Memory Requirement - Loading and Training these language models dictates a certain amount of RAM. For experimenting with smaller agents 3Gb may be sufficient, but for larger agents much more RAM may be required. If training takes more than a few minutes for small agents or more than half an hour for large agents, it is a good indicator that you are running low on memory.
- Windows Shared Drives - On Windows, make sure you have shared the drive with Docker so that it can mount directories inside of the container.
- Docker Toolbox - The below instructions work best on Linux or when using Docker for Windows or Docker for Mac.
- Elasticsearch user - if your user isn't 1000 you'll need to execute the below code to set the Elasticsearch folder to the correct ownership.
sudo chown -R 1000:1000 local-storage/elasticsearch
-
Installing Prerequisites - Make sure you have Docker, Docker Compose, and Git installed:
- Docker (If faced with any choices, get the Community Edition ("CE"))
- Docker Compose (should be auto-installed as part of Docker on mac and windows)
-
Downloading - Download the current release from this page: https://github.com/samtecspg/articulate/releases/latest
-
Extracting - Unzip and get a terminal or command prompt in the new directory
Once you've got everything extracted you're ready to start Articulate. The below instructions cover running locally and in the cloud:
Running locally or in the cloud is as simple as running a single command:
docker-compose up
sudo may be required
After Docker downloads the images and starts them Articulate should be available at http://localhost:8080 and the API documentation should be available at http://localhost:7500/documentation