Awesome RL with Human Feedback / 2021 / arXiv

Revisiting the Weaknesses of Reinforcement Learning for Neural Machine Translation

作者：Samuel Kiegeland, Julia Kreutzer · 单位：Heidelberg University, Google Research · 来源：arXiv · 发布日期：2021

奖励建模策略优化策略梯度人类反馈强化学习长文本/对话
新标签打开 AlphaXiv新标签打开 Code新标签打开 Dataset
Revisiting the Weaknesses of Reinforcement Learning for Neural Machine Translation figure
AlphaXiv 中文概览（可滚动查看）