On Sat, 2026-06-13 at 10:37 +0200, Oskar Roesler wrote:
> I've played around with a check script that uses Qwen2.5-Coder-7B-
> Q8_0-GGUF 
> locally in llama.cpp and after adding lots of grep checks for
> specific 
> assessments (aka harnesses) to stop hallucinations causing false
> positives, 
> it works surprisingly good for such a small model.


Oskar, thanks for the insight.

I have amended AURSCAN to explicitely support local LLMs and got
success reports about OpenAI.
Would you like to try with Qwen Coder? I would welcome and appreciate
any feedback if this works.

https://github.com/manticore-projects/aurscan#%EF%B8%8F-configuration

Thank you already, cheers and best
Andreas

Reply via email to