Autocomplete always produces syntax errors with qwen2.5-coder:7b (Ollama) in VS Code

[HireHoffee]

Before submitting your bug report

• I've tried using the "Ask AI" feature on the Continue docs site to see if the docs have an answer
• I'm not able to find a related conversation on GitHub discussions that reports the same bug
• I'm not able to find an open issue that reports the same bug
• I've seen the troubleshooting guide on the Continue Docs

Relevant environment info

- OS:
- Continue version:
- IDE version:
- Model:
- config:
  

  
  OR link to agent in Continue hub:

Description

When using autocomplete in VS Code with Continue.dev and qwen2.5-coder:7b via Ollama, the generated completion frequently results in syntax errors.

The model often:

• Leaves unclosed brackets ({, ()
• Duplicates closing braces (})
• e.t.c

Config:

• name: qwen2.5-coder:7b
  provider: ollama
  model: qwen2.5-coder:7b
  roles:
  • autocomplete
    defaultCompletionOptions:
    temperature: 0.3
    contextLength: 16000
    autocompleteOptions:
    useCache: true
    useImports: true
    useRecentlyEdited: true
    useRecentlyOpened: true
    maxPromptTokens: 2048
    prefixPercentage: 0.7
    maxSuffixPercentage: 0.3

Any suggestions how to fix this ?

To reproduce

No response

Log output

[operator2036]

You can always fine tune it to perform better. That is actually very optimal if you're planning to keep it offline and local. Let me know if you need help fine-tuning it.