Efficient Reinforcement Learning by Guiding Generalist World Models with Non-Curated Data
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866910949522276352 |
|---|---|
| author | Zhao, Yi Scannell, Aidan Zhao, Wenshuai Hou, Yuxin Cui, Tianyu Chen, Le Büchler, Dieter Solin, Arno Kannala, Juho Pajarinen, Joni |
| author_facet | Zhao, Yi Scannell, Aidan Zhao, Wenshuai Hou, Yuxin Cui, Tianyu Chen, Le Büchler, Dieter Solin, Arno Kannala, Juho Pajarinen, Joni |
| contents | Leveraging offline data is a promising way to improve the sample efficiency of online reinforcement learning (RL). This paper expands the pool of usable data for offline-to-online RL by leveraging abundant non-curated data that is reward-free, of mixed quality, and collected across multiple embodiments. Although learning a world model appears promising for utilizing such data, we find that naive fine-tuning fails to accelerate RL training on many tasks. Through careful investigation, we attribute this failure to the distributional shift between offline and online data during fine-tuning. To address this issue and effectively use the offline data, we propose two essential techniques: \emph{i)} experience rehearsal and \emph{ii)} execution guidance. With these modifications, the non-curated offline data substantially improves RL's sample efficiency. Under limited sample budgets, our method achieves a 102.8\% relative improvement in aggregate score over learning-from-scratch baselines across 72 visuomotor tasks spanning 6 embodiments. On challenging tasks such as locomotion and robotic manipulation, it outperforms prior methods that utilize offline data by a decent margin. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2502_19544 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Efficient Reinforcement Learning by Guiding Generalist World Models with Non-Curated Data Zhao, Yi Scannell, Aidan Zhao, Wenshuai Hou, Yuxin Cui, Tianyu Chen, Le Büchler, Dieter Solin, Arno Kannala, Juho Pajarinen, Joni Machine Learning Robotics Leveraging offline data is a promising way to improve the sample efficiency of online reinforcement learning (RL). This paper expands the pool of usable data for offline-to-online RL by leveraging abundant non-curated data that is reward-free, of mixed quality, and collected across multiple embodiments. Although learning a world model appears promising for utilizing such data, we find that naive fine-tuning fails to accelerate RL training on many tasks. Through careful investigation, we attribute this failure to the distributional shift between offline and online data during fine-tuning. To address this issue and effectively use the offline data, we propose two essential techniques: \emph{i)} experience rehearsal and \emph{ii)} execution guidance. With these modifications, the non-curated offline data substantially improves RL's sample efficiency. Under limited sample budgets, our method achieves a 102.8\% relative improvement in aggregate score over learning-from-scratch baselines across 72 visuomotor tasks spanning 6 embodiments. On challenging tasks such as locomotion and robotic manipulation, it outperforms prior methods that utilize offline data by a decent margin. |
| title | Efficient Reinforcement Learning by Guiding Generalist World Models with Non-Curated Data |
| topic | Machine Learning Robotics |
| url | https://arxiv.org/abs/2502.19544 |