OpenAI用人类反馈强化学习训练摘要模型

We’ve applied reinforcement learning from human feedback to train language models that are better at summarization.