Missing first line in generated code using VSCode and ollama

[tz1112]

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 believe this is a bug. I'll try to join the Continue Discord for questions
• 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: Ubuntu 24.04.1
- Continue version: 1.0.19 / 1.1.71
- IDE version: 1.102.3
- Model: qwen2.5-coder, tinyllama
- config:
  
name: Local Assistant
version: 1.0.0
schema: v1

models:
  - name: Qwen2.5 Coder
    provider: ollama
    model: qwen2.5-coder:7b
    apiBase: http://192.168.178.71:11434
    default: true
  - name: qwenlocalhost
    provider: ollama
    model: qwen2.5-coder:1.5b
    apiBase: http://localhost:11434
    default: true
  - name: llamalocalhost
    provider: ollama
    model: tinyllama:1.1b
    apiBase: http://localhost:11434
    default: true

Description

If i use ctrl+i to generate code, the suggested code is missing the first line ~95% of the time. The issue persists across different models, the current stable release as well as the current pre-release version. The continue console in vscode shows the correct response, with the first line as expected.

Example:
Prompt: "Generate a for loop from 0 to 100"
Output in Editor:

Output of continue console:

To reproduce

No response

Log output