Research map / Reinforcement learning / Offline RL
AdamO: A Collapse-Suppressed Optimizer for Offline RL
Nan Qiao, Sheng Yue, Shuning Wang, Ju Ren
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 |
|---|---|---|---|
| replace Adam with AdamO | TD3+BC · MuJoCo locomotion avg (10k samples) | normalized score: 29.6 -> 66.1 | Improved |
| replace Adam with AdamO | TD3+BC · AntMaze avg | normalized score: 15.0 -> 31.8 | Improved |
| replace Adam with AdamO | IQL · MuJoCo locomotion avg (10k samples) | normalized score: 47.3 -> 53.6 | Improved |
| replace Adam with AdamO | ReBRAC · Adroit avg | normalized score: 26.2 -> 25.6 | No clear effect |
| replace Adam with AdamO | PARS · AntMaze avg | normalized score: 70.6 -> 73.7 | Improved |
| replace Adam with AdamO | PARS · AntMaze-umaze | normalized score: 97.3 -> 93.5 | Hurt |
| replace Adam with AdamO | ACTIVE · Adroit avg | normalized score: 26.2 -> 27.5 | Improved |
| replace Adam with AdamO | SQOG · MuJoCo locomotion avg (10k samples) | normalized score: 53.6 -> 59.4 | Improved |
| replace Adam with AdamO | TD3+BC · Adroit avg | normalized score: 0.3 -> 20.9 | Improved |
| use AdamO instead of Kron optimizer | TD3+BC · MuJoCo locomotion avg, optimizer comparison | normalized score: 58.0 -> 66.1 | Improved |
| use AdamO instead of Kron optimizer | TD3+BC · Adroit avg, optimizer comparison | normalized score: 67.6 -> 93.4 | Improved |
| use AdamO instead of spectral normalization | TD3+BC · Walker2d-medium-expert | normalized score: 95.9 -> 98.5 | Improved |
Closest papers by meaning
- A Clean Slate for Offline Reinforcement Learning (NeurIPS 2025 · similarity 0.89)
- Optimistic Critic Reconstruction and Constrained Fine-Tuning for General Offline-to-Online RL (NeurIPS 2024 · similarity 0.89)
- Uni-O4: Unifying Online and Offline Deep Reinforcement Learning with Multi-Step On-Policy Optimization (ICLR 2024 · similarity 0.88)
- Offline-Boosted Actor-Critic: Adaptively Blending Optimal Historical Behaviors in Deep Off-Policy RL (ICML 2024 · similarity 0.88)
- Adam on Local Time: Addressing Nonstationarity in RL with Relative Adam Timesteps (NeurIPS 2024 · similarity 0.88)
- Beyond OOD State Actions: Supported Cross-Domain Offline Reinforcement Learning (AAAI 2024 · similarity 0.88)
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.
