Awesome RL with Human Feedback / 2021 / arXiv

PEBBLE: Feedback-Efficient Interactive Reinforcement Learning via Relabeling Experience and Unsupervised Pre-training

作者：Kimin Lee, Laura Smith, Pieter Abbeel · 单位：via Relabeling Experience and Unsupervised Pre-training · 来源：arXiv · 发布日期：2021

奖励建模偏好学习样本效率数据筛选
新标签打开 AlphaXiv新标签打开 Code
PEBBLE: Feedback-Efficient Interactive Reinforcement Learning via Relabeling Experience and Unsupervised Pre-training figure
AlphaXiv 中文概览（可滚动查看）