Autocomplet does not work with VSCODE and ollama

[mjjank]

Everything works fine, but not autocoplet. When I look in the continue console, I get

Here is chat promt:

Putting the prompt from the console to chat window gives the correct autocompletion text.

This is my config:

- name: Autodetect
    provider: openai
    apiBase: ####
    apiKey: ####
    model: AUTODETECT
    roles:
      - chat
      - edit
      - apply
      - rerank
      - autocomplete
    defaultCompletionOptions:
      temperature: 0.2
      maxTokens: 10000

[uinstinct]

Which model have you configured for autocomplete?

[mjjank]

I am using gtp oss 120b. Is it too large and a complex model? When I copy-paste autocomplet promt to standard chat, the output seems to be reasonable.

[mjjank]

Just to add in the VS Code console I see this:

[Extension Host] trigger reason: accepting
console.ts:137 [Extension Host] No suitable jump location found after trying all positions

[bdougie]

We have recommendations for autocomplete. Unfortunately gpt-oss:120b is not a recommendation, as speed is priority for this experience. See the link here for the recommendations https://docs.continue.dev/customization/models#recommended-models