OpenAI最新的论文：《Let’s Verify Step by Step》
OpenAI训练了一个模型，通过奖励每一个正确的推理步骤（“过程监督”），而不仅仅是奖励正确的最终结果（“结果监督”），在数学问题解决方面达到了新的最高水平。…
AK: Open AI releases paper + dataset
Let’s Verify Step by Step
trained a model to achieve a new state-of-the-art in mathematical problem solving by rewarding each correct step of reasoning (“process supervision”) instead of simply rewarding the correct final answer (“outcome…
https://twitter.com/i/web/status/1663981726647894027
