Research map / Reinforcement learning / Policy gradient and actor-critic
Zero-Shot Off-Policy Learning
Arip Asadulaev, Maksim Bobrin, Salem Lahlou, Dmitry V. Dylov, Fakhri Karray, Martin Takáč
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 FB latent inference with ZOL occupancy-ratio latent optimization | FB · ExORL DMC-state avg | return: 592.5 -> 627.2 | Improved |
| replace FB latent inference with ZOL | FB · OGBench avg | success rate: 0.45 -> 0.47 | Improved |
| ZOL vs FB on puzzle-3x3-play | FB · OGBench puzzle-3x3-play | success rate: 0.0 -> 0.14 | Improved |
| ZOL vs ReLA on antmaze-large-navigate | ReLA · OGBench antmaze-large-navigate | success rate: 0.77 -> 0.63 | Hurt |
| ZOL vs FB on antmaze-medium-navigate | FB · OGBench antmaze-medium-navigate | success rate: 0.95 -> 0.91 | Hurt |
| ZOL vs ReLA on Quadruped stand | ReLA · ExORL Quadruped stand | return: 846 -> 660 | Hurt |
| ZOL vs TD-JEPA on Quadruped walk | TD-JEPA · ExORL Quadruped walk | return: 600 -> 411 | Hurt |
| ZOL vs FB on Point Mass bottom right | FB · ExORL Point Mass maze bottom right | return: 165 -> 225 | Improved |
| use fewer inner optimization steps (ablation) | ZOL · DMC (normalized return) | normalized return | Improved |
| add small trust-region penalty (ablation) | ZOL · DMC (normalized return) | normalized return | Improved |
| increase chi2 penalty to large value (ablation) | ZOL · DMC (normalized return) | normalized return | Hurt |
| vary weight clipping value (ablation) | ZOL · DMC (normalized return) | normalized return | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- TD-JEPA — TD-JEPA: Latent-predictive Representations for Zero-Shot Reinforcement Learning (ICLR 2026) · better on 2 of 3 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Zero-Shot Reinforcement Learning from Low Quality Data (NeurIPS 2024 · similarity 0.89)
- Foundation Policies with Hilbert Representations (ICML 2024 · similarity 0.89)
- Provable Zero-Shot Generalization in Offline Reinforcement Learning (ICML 2025 · similarity 0.89)
- Distributional Successor Features Enable Zero-Shot Policy Optimization (NeurIPS 2024 · similarity 0.89)
- Zero-Shot Whole-Body Humanoid Control via Behavioral Foundation Models (ICLR 2025 · similarity 0.88)
- Towards Robust Zero-Shot Reinforcement Learning (NeurIPS 2025 · 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.
