Research map / Multi-agent systems and games / Human-AI and goal-conditioned
Recurrent Environment Simulators
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 consecutive prediction-dependent transitions instead of only observation-dependent transitions (ablation) | Atari prediction-dependent simulator · Atari games | long-term prediction accuracy | Helped |
| Use more prediction-dependent transitions rather than observation-dependent transitions (ablation) | Atari prediction-dependent simulator · Atari games | short-term prediction accuracy | Hurt |
| Alternate prediction-dependent and observation-dependent transitions rather than using consecutive prediction-dependent transitions (ablation) | Atari prediction-dependent simulator · Atari games | long-term prediction accuracy | Hurt |
| Use a prolonged observation-dependent-only training phase before prediction-dependent training (ablation) | Atari prediction-dependent simulator · Atari games | prediction accuracy | Hurt |
| Increase prediction length T while using prediction-dependent transitions (ablation) | Atari prediction-dependent simulator · Atari games | long-term prediction accuracy | Helped |
| Increase prediction length T while using prediction-dependent transitions (ablation) | Atari prediction-dependent simulator · Atari games | short-term prediction accuracy | Hurt |
| Use multiple truncated-BPTT subsequences instead of one, with 100% prediction-dependent transitions (ablation) | Atari prediction-dependent simulator · Pong and Seaquest | long-term prediction accuracy | Helped |
| Use multiple truncated-BPTT subsequences instead of one, with 0% or 33% prediction-dependent transitions (ablation) | Atari prediction-dependent simulator · Pong and Seaquest | prediction accuracy | No clear effect |
| Use multiple truncated-BPTT subsequences instead of one (ablation) | Atari prediction-dependent simulator · Riverraid | long-term prediction accuracy | Helped |
| Use only prediction-dependent transitions instead of a mixed training scheme (ablation) | Atari simulator · Breakout, Ms Pacman, Qbert, Space Invaders | overall prediction quality | Hurt |
| Use only prediction-dependent transitions instead of a mixed training scheme (ablation) | Atari simulator · Bowling, Fishing Derby, Freeway, Pong, Riverraid, Seaquest | long-term prediction accuracy | Helped |
| Train with only prediction-dependent transitions rather than a mixed scheme (ablation) | Atari simulator used interactively · Breakout, Freeway, Pong | robustness to human-chosen actions | Hurt |
Closest papers by meaning
- Learning 3D Persistent Embodied World Models (NeurIPS 2025 · similarity 0.87)
- Diffusion for World Modeling: Visual Details Matter in Atari (NeurIPS 2024 · similarity 0.86)
- Diffusion Models Are Real-Time Game Engines (ICLR 2025 · similarity 0.86)
- Learning Interactive Real-World Simulators (ICLR 2024 · similarity 0.86)
- EDELINE: Enhancing Memory in Diffusion-based World Models via Linear-Time Sequence Modeling (NeurIPS 2025 · similarity 0.86)
- Exploration-Driven Generative Interactive Environments (CVPR 2025 · similarity 0.85)
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.