Oh, by "real world" I meant "chains of thought generated by existing reasoning LLMs" (as opposed to injecting predefined CoT like was done in the experiment), n

Oh, by "real world" I meant "chains of thought generated by existing reasoning LLMs" (as opposed to injecting predefined CoT like was done in the experiment), not human thoughts.