The paper introduces Inference‑Native Zeroth‑Order (ZO) optimization, which redefines ZO as a query‑based process that can be executed directly by inference runtimes. By exposing ZO’s query semantics and using abstractions such as ProbePlan, factorized side states, and persistent subspace reuse, the method reduces state‑management cost and DRAM traffic dramatically. Experiments on large models (OPT‑13B, Qwen3‑8B) show that inference‑native steps are nearly identical to matched‑query controls while achieving significant memory savings and efficient batching.
By Zelin Li, Caiwen Ding
The paper reports that existing knowledge‑editing benchmarks cannot evaluate the scope decision—whether a stored edit applies to a query—because they are counterfactual and lack negative examples. Using the gradient‑free editor INLAY, the authors exhaustively test every router action on 1,689 queries across three datasets and find that an oracle router achieves no gain over a static policy, and abstention never wins. The authors attribute this to the structural design of the benchmarks and demonstrate that adding a missing negative condition restores some headroom and allows abstention to win.
By Aditya Pratap Singh
Every memory-based knowledge editor in the SERAC lineage depends on a scope decision: given a query, does a stored edit apply? We report that current knowledge-editing benchmarks cannot measure this d...
The paper introduces OMVV, an online multi-verifier algorithm that maintains a pool of weak verifiers with varying costs and performance. It adaptively selects a verifier each round using an online score combiner and exponential-weights routing, providing distribution-free guarantees on false-accept and false-reject rates. Experiments on reasoning benchmarks show OMVV achieves higher accuracy at lower verification cost than any single fixed verifier across different budgets.
By Erfan Hajihashemi, Yanning Shen
arXiv:2606. 08854v1 Announce Type: cross Abstract: Standard Reinforcement Learning with Verifiable Rewards (RLVR) training allocates a fixed rollout budget to every query, without regard for what each query's difficulty means for the current policy.
By Shivchander Sudalairaj, Kai Xu, Akash Srivastava, Giorgio Giannone
arXiv:2607. 17409v1 Announce Type: cross Abstract: We study the problem of sequentially evaluating a new large language model (LLM) on a fixed question set using historical performance data from prior LLMs.
By Chia-Yu Hsu, Shubhanshu Shekhar
arXiv:2606.27457v2 Announce Type: replace-cross
Abstract: Serving open-weight language models on premises forces a trade-off between accuracy and computational cost. Hosting models of different sizes...
By Yasmin Moslem, Magdalena Kacmajor, Vasudevan Nedumpozhimana, Ammar Abbas, Solmaz Panahi, David Lynch, Zhuangzhuang Nie, Alexandros Agapitos, Aleksandar Milenovic, Hongmeng Song, Yucheng Shi, Yue Pan, Patricia Buffini, John D. Kelleher
The paper investigates run‑to‑run variance in semantic query engines that use large language models (LLMs) to evaluate rows. By re‑executing benchmarks (SemBench, LRO‑Bench, UDA‑Bench) and logging 1.2 million model calls, the authors find that between 0.07% and 10.2% of rows change output at temperature 0, with some tasks showing up to 24% change. They show that plan amplification can cause a large fraction of answers to differ, that many benchmark comparisons are unresolved, and that voting and calibrated cascades reduce but do not eliminate variance, leading to new reporting and design guidelines.
By Oktie Hassanzadeh, Dharmashankar Subramanian
arXiv:2602. 10387v2 Announce Type: replace-cross Abstract: Traditional query optimization relies on cost-based optimizers that estimate execution cost (e.
By Mehmet Hamza Erol, Xiangpeng Hao, Federico Bianchi, Ciro Greco, Jacopo Tagliabue, James Zou
The paper introduces Adaptive Index Reuse (AIR), a method that dynamically determines how many layers of a sparse attention indexer should refresh during a language agent’s reasoning process. AIR replaces the static layer‑refresh schedule of IndexCache with a small controller that selects from a nested library of patterns based on recent computation signals and tool events, optimizing for task reward under a compute budget. Experiments on Qwen3‑4B‑Instruct‑2507 with a 128K context show AIR improving success rates on SWE‑bench Verified, Terminal‑Bench 2.0, and three τ²‑bench domains while reducing time per successful task, and it matches or surpasses the static schedule across 50 varied settings.
By Guangyuan Wang, Yifan Zhang, Shilong Liu
arXiv:2607. 06893v1 Announce Type: cross Abstract: The Stochastic-Oracle Turing Machine (SOTM) framework models AI-augmented computation as the interaction of a probabilistic Turing machine with an oracle whose responses are drawn from context-dependent distributions.
By Jie Wang
The paper introduces LAST-CQ, a five-agent, training‑free, execution‑grounded framework for Text‑to‑Cypher that evaluates which components of an agentic pipeline contribute most to performance. Experiments on 2,471 live‑database queries across six backbones show that removing correction reduces execution‑BLEU by 3.1–12.3%, while substituting schema‑grounded feedback with raw error strings has negligible impact. Parallel sampling degrades quality by 10–11%, whereas failure detection and retry routing recover 91.7% of initially failed queries, highlighting that simple failure handling is more effective than sophisticated feedback or increased sampling.
By Ioannis Prokopiou, Athanasios Aidinis, Panagiotis-Christos Kyrmpatsos, Pantelis Vikatos