How to run Ollama with docker

Your own LLM portable

The world of computing has always been in constant change, but it is certainly true that in the last year, since the appearance of ChatGPT, the pace of these changes has multiplied exponentially. Probably, if you are dedicated to the world of software development, either through product creation or through the provision of software services for companies, you will have heard about Ollama, which is an open-source version of LLMs (Large Language Models).

Ollama is very simple to use and allows us to run a ChatGPT-like LLM locally on our computers. The great thing is that we can select different re-trained models to be used with Ollama, which is fantastic because we can have several pre-trained models on a server for different purposes. For example, if you have a package delivery company, you could have a ChatBot with Ollama to resolve user queries and another ChatBot to manage incidents; these can be pre-trained models with very different objectives, all on our servers.

To use Ollama, we can download its binary, install it, select a model, and start operating with it. However, I will go a step further and share a repository:

https://github.com/fjpulidop/ollama

I have created where, in a parameterized way, you can select the model, execute:

make launch

, and automatically have a web service running on your computers at:

localhost:3000

We will have access to a web interface similar to chat-gpt with which we can interact.

You can find more information in the README; you just need to have Docker installed and nothing else.

Best regards.