The study compares an equal‑budget Monte Carlo Tree Search (MCTS) scheduler with a randomized‑restart list scheduler for directed‑acyclic‑graph (DAG) workflows, ensuring all methods receive the same number of candidate schedules. Across 768 validated schedules, MCTS performed 2.88% worse than the restarts at a budget of 2048 evaluations, with the deficit persisting even when the selection rule was changed to textbook UCT. The results, limited to synthetic domains with 8–32 tasks and up to 2048 evaluations, suggest that simpler restart strategies can outperform MCTS in certain workflow structures.
The paper investigates the use of test‑time search, specifically Monte Carlo tree search (MCTS), to improve agent performance in terminal‑based computer‑use environments. It finds that conventional MCTS does not significantly boost task success due to limited critic information and poor intermediate state evaluation. The authors propose a new strategy called Probe‑and‑Resume, where a critic is invoked only when the agent believes it has finished, and it is given read, write, and execute access to a forked copy of the environment to probe hidden state. This approach raises success rates by 8.4 percentage points over standard MCTS and nearly matches a more expensive Codex CLI agent.
By Jiakai Xu, Tiancheng Ge, Tujie Guan, Yusheng Li, Rohan Timmaraju, Eugene Wu, Kostis Kaffes
arXiv:2604.07799v3 Announce Type: replace-cross
Abstract: Robots deployed for long periods keep improving their skills, and each update changes a released system. We treat this as a software-lifecycl...
By Xue Qin, Simin Luan, Cong Yang, Zhijun Li
arXiv:2608. 14761v1 Announce Type: cross Abstract: At a finite public-chance cut, counterfactual regret minimization (CFR) must choose how many outcomes to evaluate before each regret update.
By Jiaxing Guo, Lei Ye
The study investigates how inference‑time interventions and weight consolidation affect open‑ended generation in an online bin‑packing task. By iteratively generating, verifying, selecting, and consolidating with LoRA, the model’s outputs shift toward higher value, reducing excess by 1.7 points and outperforming random consolidation by 3.1 points. Across three independent runs, the mean performance remained consistent, and the best candidates converged to the classic heuristic’s level without exceeding it, while consolidation also lowered the proportion of better‑than‑classic candidates but increased their absolute number.
By Roberto I. Ono Filho
arXiv:2608. 04611v1 Announce Type: cross Abstract: Frontier coding models now match or exceed strong human reference points on programming benchmarks, yet benchmark success does not imply maintainable software.
By Ruitong Li, Binjie Guo, Aisheng Mo, Guowei Su, Han Wang, Jie Li, Ru Zhang
The paper introduces NP‑Bench, a benchmark and a proactive scheduling planner that coordinates parallel large‑language‑model coding agents. By partitioning work scopes and ordering merges ahead of time, the planner improves clean‑integration rates from 1/9 to 9/9 and eliminates merge conflicts, outperforming both no‑coordination and reactive‑detection baselines. It also demonstrates that cross‑session memory can eliminate repeated mistakes and that routing facts to agents does not improve long‑context accuracy at scale.
By Sumanyu Muku
The paper introduces a formal framework for agent harnesses that guarantees termination, prevents drift, and enforces spend limits through bounded loops, gates, and repair relations. It proves that these guarantees hold even with repair budgets and demonstrates the effectiveness of the system by identifying vacuous gates and achieving low false‑accept rates in a 69‑loop catalogue. The authors provide an instrumented implementation and a held‑out mutant corpus to validate gate correctness.
By Varun Pratap Bhardwaj, Garima Singh, Arun Pratap Bhardwaj
arXiv:2608. 07911v1 Announce Type: new Abstract: Mixture-of-Experts (MoE) models have outgrown accelerator memory, and offloading expert weights to host memory is now standard.
By Yu Zhang
arXiv:2608.07911v4 Announce Type: replace
Abstract: Mixture-of-Experts (MoE) models have outgrown accelerator memory, and offloading expert weights to host memory is now standard. This makes expert c...
By Yu Zhang
The paper investigates how the ordering of training data written under two incompatible but correct conventions influences a model’s learned parameters. It shows that the learning‑rate schedule acts as an averaging operator that determines the ordering effect, with constant schedules producing larger effects than decaying ones. Experiments on a single corpus with fixed budget demonstrate a statistically significant 12.29‑sigma shift in parameter commitment, yet this shift is zero when measured with a convention‑agnostic metric.
By Wenhui Chen
PreSTO introduces Predictive SubTree Prefetching to accelerate PowerMH, a power sampling method for large language models. By prefetching future proposals in a tree of accept/reject decisions and using a predictive traversal rule, PreSTO maintains the same Metropolis–Hastings transition kernel while reducing model calls. Experiments on seven datasets and five base LLMs show median speedups of 1.71×, 1.91×, and 2.33× for PowerMH, EntropyCut, and MultiTryMH, respectively, with fewer and shorter model calls.
By Nan Jiang, Panagiotis Theodoropoulos, Wei Deng