Research map / Reinforcement learning / Offline RL
NeoRL: A Near Real-World Benchmark for Offline Reinforcement Learning
Rong-Jun Qin, Xingyuan Zhang, Songyi Gao, Xiong-Hui Chen, Zewen Li, Weinan Zhang, Yang Yu
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 |
|---|---|---|---|
| compare against deterministic policy instead of behavior policy | CQL · NeoRL 51 tasks | win ratio: 94.12% -> 88.24% | Hurt |
| compare against deterministic policy instead of behavior policy | PLAS · NeoRL 51 tasks | win ratio: 72.55% -> 60.78% | Hurt |
| use CQL instead of BC | BC · NeoRL 51 tasks | average rank: 4.63 -> 2.45 | Improved |
| use PLAS or BCQ instead of BC | BC · NeoRL 51 tasks | average rank: 4.63 -> 4.61 (PLAS), 5.63 (BCQ) | No clear effect |
| use MOPO instead of BC | BC · NeoRL 51 tasks | average rank: 4.63 -> 7.08 | Hurt |
| add KL constraint to MB-PPO (ablation) | MB-PPO · NeoRL 51 tasks | average rank: 6.29 -> 3.90 | Improved |
| use model-based methods instead of model-free | offline RL · NeoRL | episode return | Mixed |
| increase dataset size or quality | offline RL algorithms · NeoRL | episode return | Mixed |
| select policies with FQE instead of uniform random selection | FQE · Walker2d-v3 policy selection | Top-1 mean score | Hurt |
| select policies by FQE rank | FQE · Walker2d-v3 | RC score: -.005 +- .222 | No clear effect |
Closest papers by meaning
- A Clean Slate for Offline Reinforcement Learning (NeurIPS 2025 · similarity 0.93)
- Forecasting in Offline Reinforcement Learning for Non-stationary Environments (NeurIPS 2025 · similarity 0.92)
- Is Value Learning Really the Main Bottleneck in Offline RL? (NeurIPS 2024 · similarity 0.91)
- OGBench: Benchmarking Offline Goal-Conditioned RL (ICLR 2025 · similarity 0.91)
- Dispelling the Mirage of Progress in Offline MARL through Standardised Baselines and Evaluation (NeurIPS 2024 · similarity 0.91)
- Simple Ingredients for Offline Reinforcement Learning (ICML 2024 · similarity 0.91)
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.
