cannot assign model to autocomplete

[mark-hahn]

I have codestral assigned to chat and it works. When I click on Setup Autocomplete Model it just sends me to a page about autocomplete, like what model to choose, but there is nothing saying how to actually assign and use autocomplete. So I'm stuck.

[bdougie]

@mark-hahn sorry to hear you are stuck . Can you share if you are using the Local Assistant experience and not a hosted experience on the Hub. I'd be happy to update the docs after I gather more context.

But assuming you are using the local experience via the config.yaml with ollama. You can add something like this to the yaml.

models: 
  - name: Autodetect
    provider: ollama
    model: AUTODETECT
    roles:
      - chat
      - edit
      - apply
      - rerank
      - autocomplete

The docs have the example with Codestral in the docs https://docs.continue.dev/reference#models

Int he VS code extension, locally you would need to point to a provider. https://docs.continue.dev/customize/model-providers/overview

I also realize a lot of this context could be centralized to provided as a better experience, so any feedback here would be helpful.