Research map / Architectures and efficiency / RNNs and normalization
Fast and Precise: Adjusting Planning Horizon with Adaptive Subgoal Search
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 |
|---|---|---|---|
| prioritize longer subgoals and fall back to shorter distances when needed (ablation) | AdaSubS · INT, 50 nodes | success rate: 51.5% -> 59.0% | Helped |
| prioritize longer subgoals over value-ranked states alone (ablation) | adaptive subgoal search · INT, 50 nodes | success rate: 39.5% -> 59.0% | Helped |
| mix subgoal distances instead of using one fixed distance (ablation) | kSubS · INT, 50 nodes | success rate: 4.0% -> 17.0% | Helped |
| add a verifier to reject unreachable subgoals (ablation) | Longest-first · INT, 50 nodes | success rate: 51.5% -> 59.0% | Helped |
| add a verifier to reject unreachable subgoals (ablation) | Longest-first · INT, 1000 nodes | success rate: 95.5% -> 95.7% | No clear effect |
| add a verifier to reject unreachable subgoals (ablation) | Longest-first · Sokoban, 100 nodes | success rate: 63.4% -> 72.2% | Helped |
| use scheduled generator alternation rather than longest-first prioritization (ablation) | adaptive subgoal search · INT, 50 nodes | success rate: 59.0% -> 54.0% | Hurt |
| use state-value-first ordering rather than global longest-distance priority (ablation) | adaptive subgoal search · Rubik's Cube, 400 nodes | success rate: 58.0% -> 33.4% | Hurt |
| use a single fixed subgoal distance rather than adaptive distances (ablation) | AdaSubS · Sokoban, 100 nodes | success rate: 26.0% -> 72.2% | Hurt |
Closest papers by meaning
- SPIRAL: Symbolic LLM Planning via Grounded and Reflective Search (AAAI 2026 · similarity 0.86)
- Counting and Reasoning with Plans (AAAI 2025 · similarity 0.85)
- PathWise: Planning through World Model for Automated Heuristic Design via Self-Evolving LLMs (ICML 2026 · similarity 0.85)
- System 1.x: Learning to Balance Fast and Slow Planning with Language Models (ICLR 2025 · similarity 0.85)
- ACPBench: Reasoning About Action, Change, and Planning (AAAI 2025 · similarity 0.85)
- HyperTree Planning: Enhancing LLM Reasoning via Hierarchical Thinking (ICML 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.