CodeLlama - 34b - instruct via HuggingFace does not start coding.

[BlackHawk74]

Describe the bug

The model does not start coding as soon as I give it a prompt, it just simply asks more and more details, and does not ever start to write a single bit of code

Link to the Bolt URL that caused the error

http://localhost:5173/chat/6

Steps to reproduce

1. start the dev server
2. select CodeLlama - 34b - instruct via HuggingFace
3. Give a prompt

Expected behavior

expected behavior is to start writing code

Screen Recording / Screenshot

Platform

• OS: Windows, MacOS
• Browser: Chrome
• Version: latest

Provider Used

HuggingFace

Model Used

CodeLlama - 34b - instruct

Additional context

No response

[leex279]

Try with optimized prompt:

Just testet and works for me.

[BlackHawk74]

Try with optimized prompt:

Just testet and works for me.

Will test this once and get back to you.

[BlackHawk74]

Try with optimized prompt:

Just testet and works for me.

Still doesn't work, any other fixes?

[leex279]

Hmm need to test myself, but half an hour ago the a new context optimization was pushed to main branch. Did you already pulled the latest changes and tried with that?
Make sure you toggle on then "use context optimization" in settings