Awesome RL with Human Feedback / 2023 / arXiv

The Wisdom of Hindsight Makes Language Models Better Instruction Followers

作者：Tianjun Zhang, Fangchen Liu, Justin Wong, Pieter Abbeel, Joseph E. Gonzalez · 来源：arXiv · 发布日期：2023

训练稳定性策略梯度人类反馈强化学习框架/系统
新标签打开 AlphaXiv新标签打开 Code
The Wisdom of Hindsight Makes Language Models Better Instruction Followers figure
AlphaXiv 中文概览（可滚动查看）