Now supports running local models using Ollama, and using them via Gagarin. You will need to run the toolbar version. Local models are not yet supported on Windows, but have been tested on Mac and should work just fine on most linux distributions (but you may need to build your own binary for linux). Models are automatically created under the ollama provider using their local names, i.e., gemma:2b.
Here's an example config for using gemma:2b locally:
version: 0.0.1
agents:
- name: local_demo
description: My agent for demoing Fyodorov with local model
model: gemma:2b
prompt: My name is Daniel. Ask me if I'm enjoying Fyodorov AI so far.