autollama

From the ollama help discord channel:

Docker compose: How do you pull a model automatically with container creation?

The conundrum of containerizing Ollama is that it must be running to pull a model. If you run ollama using docker compose, it doesn't provide a way to pull a model from the ollama registry. What's a budding generative AI nerd to do?

The answer is more docker compose, as in docker compose exec, which runs a script to download a model. …

compose discord docker generative nerd ollama registry running