Research map / Reinforcement learning / Offline RL
RL Unplugged: A Suite of Benchmarks for Offline Reinforcement Learning
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use behavior cloning rather than value-based offline RL on difficult, high-dimensional control tasks | Offline RL baselines · DM Control Suite — Humanoid run | episodic reward: BC: 382$\pm$3; RABM: 303$\pm$6; D4PG: 1.72$\pm$1.66 | Mixed |
| Use behavior cloning or RABM on high-dimensional, difficult-exploration locomotion tasks rather than D4PG | Offline RL baselines · DM Locomotion | episodic reward: Humanoid corridor: 220$\pm$194 (BC), 64.5$\pm$3.8 (RABM), 4.39$\pm$4.15 (D4PG) | Helped |
| Choose hyperparameters on online policy-selection tasks and transfer them to offline policy-selection tasks | Offline RL baselines · Atari 2600 | normalized score | Mixed |
| Use robust off-policy methods such as REM or IQN instead of DQN for Atari offline policy selection | Atari offline RL · Atari 2600 | normalized score | Helped |
| Use 16 τ samples instead of other tested sample counts for IQN (ablation) | IQN · Atari 2600 online policy-selection games | performance | Helped |
Closest papers by meaning
- A Clean Slate for Offline Reinforcement Learning (NeurIPS 2025 · similarity 0.93)
- ODRL: A Benchmark for Off-Dynamics Reinforcement Learning (NeurIPS 2024 · similarity 0.92)
- Is Value Learning Really the Main Bottleneck in Offline RL? (NeurIPS 2024 · similarity 0.91)
- Horizon Reduction Makes RL Scalable (NeurIPS 2025 · similarity 0.91)
- Beyond OOD State Actions: Supported Cross-Domain Offline Reinforcement Learning (AAAI 2024 · similarity 0.91)
- OGBench: Benchmarking Offline Goal-Conditioned RL (ICLR 2025 · similarity 0.90)
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.