Applying Model-agnostic DeepSeek R1-level reasoning onto any LLMgithub.com/jacobbergdahl1 pointjacob_bergdahla year ago