arXiv:2607. 11197v1 Announce Type: new Abstract: When LLMs exhibit uneven performance across planning tasks, these gaps are often attributed to task difficulty.
Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv AI.
When LLMs exhibit uneven performance across planning tasks, these gaps are often attributed to task difficulty. We argue that this explanation is incomplete, as task-level variation may reflect distinct latent planning competencies rather than differences along a single ability spectrum.
arXiv:2610. 26964v1 Announce Type: new Abstract: Large language models (LLMs) struggle with classical planning, but benchmarks rarely distinguish structural difficulty from the difficulty of interpreting symbolic vocabulary.
The paper introduces a verifiable framework for measuring the innate planning efficiency of large language models (LLMs) on sliding puzzles, focusing on complete plans generated without external tools. It defines a Success-Weighted Plan Efficiency (SPE) metric as the product of execution reliability, planning coordination, and action economy, and validates these factors through trajectory replay and certified optimality. Experiments on 4×4 puzzles show that models can finish all trials yet become less economical over longer horizons, and that high efficiency among completed plans can coexist with poor goal attainment, highlighting distinct coordination and completion issues.
The paper introduces the Dual-Latent World Model (Dual-WM), which separates local execution and long-range planning into distinct latent spaces and dynamics models. A new learning method, Long-Horizon Representation Learning with Weighted Rollout (LoRe), supervises predictions at both levels using exponential horizon weights. Experiments on five goal-conditioned visual control tasks show that Dual-WM improves success rates over strong baselines, especially at longer horizons.
arXiv:2510.11620v3 Announce Type: replace Abstract: Test-time scaling seeks to improve reasoning through additional inference-time computation, commonly using solution sampling with majority voting t...
GRASP is a multi-stage planning framework that improves the reliability of large language models on complex tasks. It separates planning into three specialized modules—GenPlan for global macro-guidelines, RevPlan for exploring localized strategies, and VerPlan for multi-criteria evaluation—allowing context isolation and strict macro-regularization. Experiments show GRASP outperforms direct LLM planners by significant margins on datasets such as Natural Plan Calendar Scheduling, ZebraLogic, and SciBench Math, and it mitigates performance collapse in multi-task and dual-task settings.