Custom Generative AI Config instructions are false

[dhdaines]

A few issues with the documentation here: https://docs.onyx.app/gen_ai_configs/ollama

1. You need to add a model under "Model Names". This is not mentioned in the documentation.
2. What you need to add as a model name is not what the UI claims it to be. One would be inclined to think that it's something like ollama/llama3.2 (except that oops that page you link to doesn't even mention llama3.2...) but no it is just llama3.2.
3. When deploying with docker-compose you also need to make sure ollama is listening on the appropriate interface so that it can be seen from inside the docker containers. Not sure it does this by default (hopefully not as it is a gaping security hole). You can do it by starting ollama manually with OLLAMA_HOST=172.17.0.1:11434 ./bin/ollama serve for instance.

Looking forward to actually using Onyx now ;-)