Use Ollama with the official Python library

Ollama is a great way to get started with AI by using open-source and publically available large-language models locally on your computer. I wrote previously about how to get started with the experimental OpenAI API, but Ollama has a dedicated Python library that is even simpler.

1. Install the library: pip3 install ollama

2. Create a new python file: touch completion.py

3. Add the following code to completion.py:

   import ollama
   
   def get_completion(prompt, model):
       ollama.chat(model, messages=[{
           'role': 'user',
           'content': prompt,
       }])
   
       return response['message']['content']
   
   prompt = "What is the chief end of man?"
   
   print(get_completion(prompt, "mistral"))
   

4. Run the file: python3 completion.py

There is more information available in the library repo on GitHub, including examples for streaming responses and a custom client. For even more documentation on Ollama, check out the /docs directory in the main repo.