Paper without code LLM reads a paper and produce a working prototype for papers without code! No Llama-index, no LangChain, just get back to the root of the idea.