Research map / Reinforcement learning / Offline RL
MAGE: Multi-scale Autoregressive Generation for Offline Reinforcement Learning
Chenxing Lin, Xinhui Gao, Haipeng Zhang, Xinran Li, Haitao Wang, Songzhu Mei, Chenglu Wen, Weiquan Liu, Siqi Shen, Cheng Wang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add multi-scale modeling: K=1 to K=8 (ablation) | MAGE · Adroit Pen-Expert | normalized score: 123.5 -> 147.8 | Improved |
| increase temporal scales from K=8 to K=10 (ablation) | MAGE · Adroit Door-Cloned | normalized score: 20.5 -> 17.0 | Hurt |
| increase temporal scales from K=8 to K=10 (ablation) | MAGE · Adroit Pen-Expert | normalized score: 147.8 -> 149.9 | No clear effect |
| model returns and states (R,S) instead of states only (ablation) | MAGE · Adroit Pen-Expert | normalized score: 127.7 -> 147.8 | Improved |
| model returns and states instead of returns, states and actions (ablation) | MAGE · Adroit Door-Cloned | normalized score: 17.2 -> 20.5 | Improved |
| model returns and states instead of actions with CQL (ablation) | MAGE · Adroit Door-Cloned | normalized score: 4.9 -> 20.5 | Improved |
| add RTG conditioning in the autoencoder (ablation) | MAGE · Adroit Pen-Expert | normalized score: 140.3 -> 147.8 | Improved |
| add RTG conditioning in the autoencoder (ablation) | MAGE · Adroit Door-Cloned | normalized score: 12.3 -> 20.5 | Improved |
| add RTG conditioning in transformer for finer scales (ablation) | MAGE · Adroit Door-Cloned | normalized score: 16.3 -> 20.5 | Improved |
| add RTG in conditioning loss (ablation) | MAGE · Adroit Pen-Expert | normalized score: 139.9 -> 147.8 | Improved |
| replace hierarchical diffusion with MAGE | HD · Adroit (mean all settings) | normalized score: 40.3 -> 66.9 | Improved |
| replace HD with MAGE | HD · Franka Kitchen | average score: 72.5 -> 88.8 | Improved |
Closest papers by meaning
- Stitching Sub-trajectories with Conditional Diffusion Model for Goal-Conditioned Offline RL (AAAI 2024 · similarity 0.90)
- Temporal Distance-aware Transition Augmentation for Offline Model-based Reinforcement Learning (ICML 2025 · similarity 0.90)
- Reward-Consistent Dynamics Models are Strongly Generalizable for Offline Reinforcement Learning (ICLR 2024 · similarity 0.89)
- Diffusion Subgoal Planning for Long-Horizon Offline Goal-Conditioned Reinforcement Learning (NeurIPS 2026 · similarity 0.89)
- Flattening Hierarchies with Policy Bootstrapping (NeurIPS 2025 · similarity 0.89)
- GTA: Generative Trajectory Augmentation with Guidance for Offline Reinforcement Learning (NeurIPS 2024 · similarity 0.89)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
