Dethinker: Strip 'thinking' content from Ollama responses
I try to use Local LLMs whenever I can, pretty much always via Ollama. Recently with qwen3 I have been really impressed with the quality of the responses. However, I kept running into the problem...