Wei-Chieh Huang, Weizhi Zhang, Yuchen Wu, Yankai Chen, Eric Hanchen Jiang, Wooseong Yang, Yiwei Yang, Henry Peng Zou 等 15 位作者
Memory is becoming core infrastructure for long-horizon LLM agents, yet existing evaluations offer limited guidance on which memory substrate, namely the underlying medium in which memory is represented and stored, should be used under different operating regimes. We introduce a unified suite of 26 performance and efficiency metrics, including auxiliary-LLM costs attributed to the write, read, and management stages, and apply it identically to 11 memory methods in a controlled harness that varies only the memory substrate. The methods cover dense and sparse indices, text records, structural stores, hierarchical stores, refinement-based memories, parametric updates, and activation-compatible context mechanisms, and we evaluate them with three backbone models on four benchmark suites spanning user-centric question answering and agent-centric decision-making. Our results show that no single substrate consistently dominates: broad retrieval benefits long-context factual QA, while excessive retrieval can harm sequential decision-making by shifting attention away from action-critical context. Scalability introduces a further routing axis, as substrates that perform well at moderate history lengths can become costly or brittle at longer horizons. These regime-dependent reversals make substrate selection a design axis. Choosing a substrate per observable operating regime improves over the best fixed substrate on our benchmarks, while query-level routing leaves most of the oracle headroom open. The findings provide empirical guidance for designing efficient, reliable, and regime-aware long-term memory for LLM agents.
Accurate prediction of surgical duration is pivotal for hospital resource management. Although recent supervised learning approaches, from machine learning (ML) to fine-tuned large language models (LLMs), have shown strong performance, they remain constrained by the need for high-quality labeled data and computationally intensive training. In contrast, zero-shot LLM inference offers a promising training-free alternative, but it lacks grounding in institution-specific clinical evidence, making its predictions clinically misaligned and prone to instability. To address these limitations, we present PREBA, a retrieval-augmented framework that integrates PCA-weighted retrieval and Bayesian averaging to ground LLM reasoning in institution-specific evidence without requiring task-specific fine-tuning. Specifically, PREBA first encodes heterogeneous clinical features into a unified representation. It then performs PCA-weighted retrieval, weighting each dimension by its contribution to the variation across institutional cases, instead of treating them as equally informative, and conditions the LLM to generate multiple candidate predictions. Finally, Bayesian averaging consolidates these predictions with population-level statistical priors, yielding a calibrated estimate that remains stable and clinically plausible. We evaluate PREBA on four real-world datasets spanning different institutions and languages, using three state-of-the-art LLMs, including Qwen3, DeepSeek-R1, and HuatuoGPT-o1. PREBA consistently outperforms zero-shot inference, reducing MAE by up to 64.9% and raising R2 to as high as 0.63. It further attains accuracy competitive with, and on several datasets superior to, fully supervised ML methods, while offering greater interpretability through retrieved exemplars and explicit priors. The code is available at https://github.com/kanxueli/PREBA.
Jun Liu, Dajie Sun, Zhenglun Kong, Zhenjiang Fan, Peiyan Dong, Fan Yang, Wenbin Zhang, Geng Yuan 等 11 位作者
World-action models increasingly apply test-time scaling, spending extra inference computation on candidate selection and adaptive allocation to improve control. Such gains are credible only when candidates carry exploitable signal, nominal compute savings become real speedup, and evaluation stays reliable. We audit all three on Cosmos Policy and RoboCasa with a premise-first, pre-registered methodology. On clean data, a learned value selector is 1.66 pp worse than random; a hindsight oracle gains only 1.40 pp, with an episode-level interval spanning zero; and 82% of states show identical candidate outcomes, leaving little selection headroom. Token pruning cuts per-forward FLOPs by up to 47% but yields at most a measured 1.15$\times$ end-to-end speedup on the deployed path (1.25$\times$ on a decode-excluded path), short of a 1.5$\times$ target, and our adaptive-allocation experiment is inconclusive under a power gate. We further identify and fix a silent rendering corruption whose rate rises sharply along compute-heavy planning paths while leaving the official direct-policy result unaffected, and release a detector, fix, and scripts. These findings show why test-time scaling should be audited as a chain of premises (decision signal, realized efficiency, evaluationintegrity) rather than by success rate alone. Conclusions are scoped to the tested configuration.
Third-party Agent Skills extend what an autonomous agent can do, but a malicious Skill can hide data exposure or output manipulation inside a legitimate workflow. Neither reading a Skill's description nor granting it tool permissions tells whether its execution meets the user's security requirements. We introduce RuntimeTickets, a protocol that enforces and audits these requirements while a Skill runs. From released malicious-Skill datasets we derive a risk model and build adversarial-benign Skill pairs. From the task's delivery contract, a language model proposes security obligations for that task. Each obligation cites the Skill lines it answers, and the ticket tightens the host's approvals into fine-grained allowed actions but never loosens them. A runtime gate checks the recipient, object and amount of every outbound send before forwarding it, and trusted monitors record the effects within their observation scope. The ticket links these checks and observations to the approved obligations of each Skill run, and a deterministic decoder turns the ticket's status into governable data flows and verifies each obligation. Neither enforcement nor decoding involves a language model, so text injected by a Skill can change the proposal but never the rule that decides a send. We evaluate RuntimeTickets at the network egress boundary of three unmodified agent frameworks against production Skill scanners and runtime baselines, among them an AgentBound-derived allow-list and SkillGuard. RuntimeTickets eliminated both failure modes of current Skill defenses, overlook (malicious Skills let through) and overkill (benign Skills blocked): its overlook rate was 0.0% and its overkill rate 0.0%, while the scanners erred on both sides. RuntimeTickets ties approved requirements to verifiable execution evidence and offers a common foundation for the safe and auditable deployment of third-party Skills.
Text-guided Video Temporal Grounding (VTG) localizes moments in an untrimmed video that correspond to a text query. Supervised approaches require costly temporal annotations and task-specific training, yet can generalize poorly under distribution shift. To reduce this dependence on task-specific supervision, training-free approaches use pretrained vision-language models to localize moments by measuring the similarity between video and query features. However, independently encoded frames provide limited temporal context, while a fixed query embedding can assign similar scores to different candidate moments within the video. To address these limitations, we propose DSE-VTG, a \underline{D}ual-\underline{S}ide \underline{E}nhancement framework that combines visual evidence across temporal granularities with video-specific query adaptation, without task-specific training. On the visual side, Multi-scale Similarity Fusion (MSF) combines frame- and clip-level similarities into a unified, temporally aware similarity profile. On the query side, Query-level Test-Time Adaptation (Q-TTA) uses pseudo-labels to optimize a lightweight additive offset for more discriminative matching within the video, without fine-tuning the backbone or calling external large language models. Experiments on three standard and two out-of-distribution (OOD) benchmarks show gains over the prior training-free methods. On Charades-STA, DSE-VTG improves mIoU by 5.89 percentage points over the strongest compared baseline under the same backbone. Under distribution shift, DSE-VTG reaches 50.86 mIoU on Charades-CG Novel-Word, surpassing the strongest supervised baseline by 2.76 mIoU. Our code will be released upon acceptance.
Jun Liu, Zhenglun Kong, Changdi Yang, Fan Yang, Tianqi Li, Peiyan Dong, Joannah Nanjekye, Hao Tang 等 15 位作者
Large language model (LLM) agents are increasingly organized into multi-agent systems for complex reasoning and decision-making tasks. However, scaling such systems can substantially increase inference cost, especially when multiple agents repeatedly process overlapping or role-irrelevant context. Many existing multi-agent frameworks focus on orchestration and role specialization, but still rely on full-context sharing or static context routing, leading to redundant token consumption, increased runtime, and inefficient memory utilization. We present \emph{Role-Conditioned Context Routing (RCR)}, an execution-time layer that allocates role-relevant shared-memory items to agents under token budgets. Given a structured shared memory and a set of role-specialized agents, RCR selectively routes task-relevant memory items to each agent according to its role, optional task-stage metadata, and token budget. RCR integrates role-conditioned relevance scoring, token-budgeted memory selection, and iterative memory updates to reduce redundant context exposure during multi-agent execution. Experiments on multi-hop QA, embodied reasoning (ALFWorld), and web-based decision-making (WebShop) show that RCR reduces token usage by 35\% on average and runtime by 38\% on average relative to Full-Context, while maintaining or improving task performance.
Generative artificial intelligence (GenAI) is increasingly embedded in university learning, yet effective human-GenAI interaction depends not only on access to these tools but also on students' ability to formulate, evaluate, and refine prompts for academic tasks. This study examined university students' perceived GenAI prompting capability and reported prompting strategies. Questionnaire data from 234 students were analyzed using descriptive statistics, measurement modelling, scenario comparisons, structural associations, and qualitative coding of optional comments. Students reported using GenAI across diverse academic activities and generally expressed confidence in their prompting capabilities, particularly in providing relevant context and refining prompts following initial outputs. However, prompting strategies showed limited sensitivity to task types: specifying output format was more strongly endorsed in a creative workshop preparation scenario than in a research-oriented assignment, whereas other matched strategies showed no adjusted differences. Students also commonly endorsed both concise instructions and providing all relevant information at the outset, suggesting that effective brevity may depend more on selecting and organizing task-relevant information than on prompt length alone. By contrast, fixed prompt templates received lower endorsement than other prompting strategies. In the adopted factor model, perceived cognition showed the clearest unique association with perceived prompting capability, while open-ended responses highlighted prompt-intent alignment as a common challenge.
Large language model agents are increasingly used for scientific discovery. However, existing benchmarks mix law inference with programming, score data fitting instead of prediction on unseen inputs, and do not separate empirical evidence from pretrained priors. We introduce SciGame, a controlled benchmark that converts scientific literature into structured scientific cards. Each card yields a pair of Original and dimensionally consistent Counterfactual environments with the same task interface. SciGame contains a Historical collection of canonical laws and a live Temporal collection curated from papers published after model knowledge cutoffs. It evaluates agents in candidate-guided Identification and open-ended Discovery settings. Each setting runs under Prior-only, Fixed, and Adaptive protocols. Experiments on five frontier models yield three findings. First, familiar priors hurt performance once the governing relation is altered. Second, adaptive experimentation helps the strongest models but harms weaker ones under limited budgets. Third, good explanations can hide quantitative failures. Recovering the exact symbolic form remains the main bottleneck, even when agents interpret evidence correctly. Current agents can investigate systems actively, but inducing exact laws in unfamiliar settings remains an open challenge.
A numerical claim generated by a financial analyst agent can match a number in a company filing but refer to the wrong entity, concept, period, unit or scope. A check based on reference answers can mark such an answer as wrong but does not explain why it fails. In this paper, we present NumTrace, a new benchmark that audits the provenance of each numerical claim made by financial analyst agents: its source and financial semantics. NumTrace currently has 540 questions over 2,163 SEC EDGAR documents. To the best of our knowledge, it is the first numerical benchmark about who holds a company's shares (Form 13F) and what its executives are owed (Item 402(i)). On a frozen snapshot of these documents, we build the reference answers according to accounting rules. NumTrace then gives each claim three separate verdicts: value, binding and support. By retaining separate value and binding verdicts, NumTrace detects right number, wrong account (RNWA) errors: a claim has an RNWA error when it passes the value check but fails the binding check. The RNWA rate quantifies the gap between value-only and joint value-and-binding pass rates. Finally, we audit three agents, Claude Opus 5.5, GPT-6 Astra and DeepSeek-V4.1-Flash. RNWA errors make up 67.5-81.8% of each agent's failing claims. In particular, on 13F co-filing group questions, about 91% of claims pass the value check, but fewer than half pass the binding check.
Fan Zhang, Mingzi Song Yokohama National University, Tokyo Institute of Technology, Rania Elbadry, Yankai Chen Cornell University, Shaobo Wang, Yixi Zhou, Xunwen Zheng, Yueru He Columbia University 等 22 位作者
Financial reporting systems increasingly leverage Large Language Models (LLMs) to extract and summarize corporate disclosures. However, most existing approaches assume a single-market setting and overlook structural differences across jurisdictions. Variations in accounting taxonomies, tagging infrastructures (e.g., XBRL vs. PDF), and aggregation conventions introduce substantial challenges for semantic alignment and reliable verification. Here, we aim to bridge this gap. We present FinReporting, an agentic workflow for localized cross-jurisdiction financial reporting. The system constructs a unified canonical ontology spanning the income statement, balance sheet, and cash flow statement, and decomposes reporting into auditable stages, including filing acquisition, extraction, canonical mapping, and anomaly logging. Rather than treating LLMs as free-form generators, FinReporting employs them as constrained verifiers operating under explicit decision rules with evidence grounding.Evaluated on annual filings from the USA, Japan, and China, FinReporting improves consistency and reliability under heterogeneous reporting regimes. We further release an interactive demo that enables cross-market inspection and supports structured export of localized financial statements. Our demo is available at https://huggingface.co/spaces/BoomQ/FinReporting-Demo. A video describing our system is available at https://www.youtube.com/watch?v=f65jdEL31Kk.
尚无 AI 解读，展示原文摘要
cs.CL · cs.AI
Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 3: System Demonstrations) · 2026 · 正式发表
Junyi Zhou City University of Hong Kong, Qiyuan Zhang City University of Hong Kong, Yufei Wang Huawei Technologies Ltd., Fuyuan Lyu Vatic Labs, Yidong Ming, Can Xu Tencent Hunyuan, Qingfeng Sun, Kai Zheng Microsoft 等 11 位作者
As Large Language Model (LLM) alignment evolves from simple completions to complex, highly sophisticated generation, Reward Models are increasingly shifting toward rubric-guided evaluation to mitigate surface-level biases. However, the community lacks a unified benchmark to assess this evaluation paradigm, as existing benchmarks lack both the discriminative complexity and the ground-truth rubric annotations required for rigorous analysis. To bridge this gap, we introduce RubricBench, a curated benchmark with 1,147 pairwise comparisons specifically designed to assess the reliability of rubric-based evaluation. Our construction employs a multi-dimensional filtration pipeline to target hard samples featuring nuanced input complexity and misleading surface bias, augmenting each with expert-annotated, atomic rubrics derived strictly from instructions. Comprehensive experiments reveal a substantial capability gap between human-annotated and model-generated rubrics, indicating that even state-of-the-art models struggle to autonomously specify valid evaluation criteria, lagging considerably behind human-guided performance.
尚无 AI 解读，展示原文摘要
cs.CL
Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · 2026 · 正式发表
Linfeng Du, Ye Yuan McGill University, McGill University and Mila - Quebec Artificial Intelligence Institute, Zichen Zhao, Fuyuan Lyu Vatic Labs, Emiliano Penaloza, Xiuying Chen Mohamed bin Zayed University of Artificial Intelligence, Zipeng Sun, Jikun Kang SalesForce.com 等 11 位作者
Large language models (LLMs) excel at general-purpose tasks, yet adapting their responses to individual users remains challenging. Retrieval augmentation provides a lightweight alternative to fine-tuning by conditioning LLMs on user history records, and existing approaches typically select these records based on semantic relevance. We argue that relevance serves as an unreliable proxy for utility: a record may be semantically similar to a query yet fail to improve generation quality or even degrade it due to redundancy or conflicting information. To bridge this gap, we propose PURPLE, a contextual bandit framework that oPtimizes UseR Profiles for LLM pErsonalization. In contrast to a greedy selection of the most relevant records, PURPLE treats profile construction as an order-sensitive generation process and utilizes a Plackett-Luce ranking model to capture complex inter-record dependencies. By training with semantically rich feedback provided by the likelihood of the reference response, our method aligns retrieval directly with generation quality. Extensive experiments on nine personalization tasks demonstrate that PURPLE consistently outperforms strong heuristic and retrieval-augmented baselines in both effectiveness and efficiency, establishing a principled and scalable solution for optimizing user profiles.
尚无 AI 解读，展示原文摘要
cs.CL
Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · 2026 · 正式发表
Zhaofen Wu, Hanrong Zhang University of Illinois at Chicago, Fulin Lin Zhejiang University, Wujiang Xu, Xinran Xu, Yankai Chen Cornell University, Henry Peng Zou University of Illinois at Chicago, Shaowen Chen 等 12 位作者
To sustain coherent long-term interactions, Large Language Model (LLM) agents must navigate the tension between acquiring new information and retaining prior knowledge. Current unified stream-based memory systems facilitate context updates but remain vulnerable to interference from transient noise. Conversely, discrete structured memory architectures provide robust knowledge retention but often struggle to adapt to fluid narrative evolution. To address this, we propose GAM, a hierarchical Graph-based Agentic Memory framework that explicitly decouples memory encoding from consolidation to effectively resolve the conflict between rapid context perception and stable knowledge retention. By isolating ongoing dialogue in a event progression graph and integrating it into a topic associative network only upon semantic shifts, our approach minimizes interference while preserving long-term consistency. Additionally, we introduce a Graph-guided, Multi-factor Retrieval strategy to enhance context precision. Experiments on LoCoMo and LongDialQA benchmarks indicate that our method consistently outperforms state-of-the-art baselines in both reasoning accuracy and computational efficiency.
尚无 AI 解读，展示原文摘要
cs.CL · cs.AI
Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · 2026 · 正式发表
Wentao Hu, Yanbo Zhai, Xiaohui Hu, Mingkuan Zhao, Shanhong yu, Xue Liu, Kaidong Yu Insitute of Artificial Intelligence, China Telecom, Shuangyong Song 等 9 位作者
Sparse Mixture-of-Experts (MoE) models have achieved remarkable scalability, yet they remain vulnerable to hallucinations, particularly when processing long-tail knowledge. We identify that this fragility stems from static Top-k routing: routers tend to favor high-frequency patterns over rare factual associations. Consequently, “specialist experts” possessing critical long-tail knowledge are often assigned low gating scores and remain “dormant”—under-prioritized for specific tokens despite their proven causal importance on other inputs. To address this, we propose Counterfactual Routing (CoR), a training-free inference framework designed to awaken these dormant experts. CoR integrates layer-wise perturbation analysis with the Counterfactual Expert Impact (CEI) metric to dynamically shift computational resources from syntax-dominant to knowledge-intensive layers while maintaining a constant total activation count, effectively retrieving causally decisive experts via virtual ablation. Extensive experiments on TruthfulQA, FACTOR, and TriviaQA demonstrate that CoR improves factual accuracy by 3.1% on average without increasing the inference budget, establishing a superior Pareto frontier compared to static scaling strategies.
尚无 AI 解读，展示原文摘要
cs.CL
Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · 2026 · 正式发表
Weixu Zhang McGill University and Mila - Quebec Artificial Intelligence Institute, Ye Yuan McGill University, McGill University and Mila - Quebec Artificial Intelligence Institute, Changjiang Han, Yuxing Tian Université de Montréal, Zipeng Sun, Linfeng Du, Jikun Kang SalesForce.com, Hong Kang 等 10 位作者
Large Language Models (LLMs) exhibit strong implicit personalization ability, yet most existing approaches treat this behavior as a black box, relying on prompt engineering or fine tuning on user data. In this work, we adopt a mechanistic interpretability perspective and hypothesize the existence of a sparse set of Preference Heads, attention heads that encode user specific stylistic and topical preferences and exert a causal influence on generation. We introduce Differential Preference Steering (DPS), a training free framework that (1) identifies Preference Heads through causal masking analysis and (2) leverages them for controllable and interpretable personalization at inference time. DPS computes a Preference Contribution Score (PCS) for each attention head, directly measuring its causal impact on user aligned outputs. During decoding, we contrast model predictions with and without Preference Heads, amplifying the difference between personalized and generic logits to selectively strengthen preference aligned continuations. Experiments on widely used personalization benchmarks across multiple LLMs demonstrate consistent gains in personalization fidelity while preserving content coherence and low computational overhead. Beyond empirical improvements, DPS provides a mechanistic explanation of where and how personalization emerges within transformer architectures.
尚无 AI 解读，展示原文摘要
cs.CL
Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · 2026 · 正式发表
Recent progress in large language models (LLMs) is largely driven by scaling training compute through either pre-training with next-token prediction (NTP) or post-training with reinforcement learning (RL). The former contributes to learning broad knowledge and skills from general data, while struggling with data inefficiency and catastrophic forgetting in continual learning settings. The latter incentivizes reasoning capabilities with strong generalization, but is constrained by limited data availability due to its reliance on human annotation. To alleviate these issues, we propose Reinforcement Learning on Pre-Training data (RLPT), which combines the advantages of learning from general data and RL. In particular, RLPT derives reward signals directly from general text data through a next-segment reasoning objective, rewarding the policy for correctly predicting next text segments conditioned on the prefix text. Experiments across multiple benchmarks and models demonstrate the effectiveness of . For example, RLPT yields substantial improvements in continual pre-training (+4.6\%) and provides a strong foundation for post-training (+3.4\%) on Qwen3-8B-Base.
尚无 AI 解读，展示原文摘要
cs.CL
Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · 2026 · 正式发表
Hong Huang City University of Hong Kong, Decheng Wu, Qiangqiang Hu Harbin Institute of Technology, Guanghua Yu Tencent, Jinhai Yang City University of Hong Kong, Jianchen Zhu, Xue Liu Mohamed bin Zayed University of Artificial Intelligence and McGill University, Dapeng Wu City University of Hong Kong
The deployment of Large Language Models (LLMs) on resource-constrained edge devices is increasingly hindered by prohibitive memory and computational requirements. While ternary quantization offers a compelling solution by reducing weights to -1, 0, +1, current implementations suffer from a fundamental misalignment with commodity hardware. Most existing methods must choose between 2-bit aligned packing, which incurs significant bit wastage, or 1.67-bit irregular packing, which degrades inference speed. To resolve this tension, we propose Sherry, a hardware-efficient ternary quantization framework. Sherry introduces a 3:4 fine-grained sparsity that achieves a regularized 1.25-bit width by packing blocks of four weights into five bits, restoring power-of-two alignment. Furthermore, we identify weight trapping issue in sparse ternary training, which leads to representational collapse. To address this, Sherry introduces Arenas, an annealing residual synapse mechanism that maintains representational diversity during training. Empirical evaluations on LLaMA-3.2 across five benchmarks demonstrate that Sherry matches state-of-the-art ternary performance while significantly reducing model size. Notably, on an Intel i7-14700HX CPU, our 1B model achieves zero accuracy loss compared to SOTA baselines while providing 25% bit savings and 10% speed up. The code is available at https://github.com/Tencent/AngelSlim.
尚无 AI 解读，展示原文摘要
cs.CL
Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · 2026 · 正式发表