偷不走推理，就自己编一个
前沿模型厂商把自家模型的详细推理过程藏了起来，只给答案。这很讽刺：他们自己就是靠蒸馏人类知识起家的，却禁止别人蒸馏他们的输出。有人靠跨模型重放加密块硬把推理挖出来，但那是违约，做不成产品。这篇换了个思路：不偷了，自己编。拿上下文和老师的答案当输入，用学生模型自家风格的推理痕迹去训练一个生成器，再配一个判别器分辨真假。结果很有意思：编出来的推理痕迹，真实度评分(0.68)和真痕迹(0.66)几乎一样；用判别器从四个候选中挑一个，评分能到 0.82。在编程任务上，用这套自产推理训练的模型，HumanEval+ 涨了 3.7 分，MBPP+ 涨了 7.1 分，而且优势在每个检查点都保持。他们怀疑原因是：用学生自己家族的风格写推理，它学起来更容易。这不是你明天能用的工具，但它指向一个趋势：当头部模型把推理过程锁进黑箱，开源社区正在用「伪造风格」绕过封锁——蒸馏的战争从内容转向了风格。
📄 原文摘要(英文)
Using an LLM to generate training data for another is among the most general ways to train one, and it remains common practice. Some frontier teams have closed that route for their own models: what comes back is an answer, or a short summary, without the detailed reasoning trace behind it. That is a little ironic: they were themselves built on distilling human knowledge, and distilling an open-source model is still entirely routine; the restriction applies to their own outputs alone. Recent work recovers the hidden reasoning by replaying encrypted blocks across models. It somehow works, but it breaks the providers' terms of service, so it is not something a product can be built on long-term. We built the alternative. We take the context and the teacher's answer as input, and generate the reasoning trace ourselves, in the style of the student's own model family. Training is two-stage, supervised and then adversarial: a generator learns to regenerate the family's own reasoning traces from their context and answer, and a critic learns how real a reasoning trace looks. Applied to a teacher's answer, the generator writes one in the family's style. An independent judge rates how likely traces are to be genuine on a scale from 0 to 1: ours average 0.68, real traces 0.66, and traces selected by the critic from four candidates 0.82. These are authenticity scores, not correctness rates. We test the idea on five benchmarks with everything else held fixed. On general tasks the five training corpora are about the same; on coding, the generated reasoning beats Fable 5's own thinking content. With the same 5,000 questions and the same teacher answers, HumanEval+ improves by 3.7 points and MBPP+ by 7.1 at the final checkpoint, and the MBPP+ lead holds at every checkpoint. We suspect that reasoning written in the student's family style is easier for it to learn from.