User reports severe looping with Qwen3.6 35B and Ornith V1

A Reddit user is experiencing persistent text generation loops when using the Qwen3.6 35B and Ornith V1 models via Copilot Chat. The issue occurs despite using specific parameters such as temperature 0.6, top_p 0.95, and a repetition penalty of 1.1.

• The user employs Q6 quantization for the models with K/V at Q8.
• Context windows are set to 128k, though only about 30k is utilized during the errors.
• The looping manifests as repetitive internal monologue, such as repeatedly checking for a missed useEffect in a specific file.

The user seeks diagnostic information to resolve the problem and stop the constant repetition.