We introduce LLM4Robot-Bench, a hierarchical, paired Sim-Real benchmark for evaluating physical interaction capabilities in contact-rich manipulation. Its 16 household tasks span three cumulative levels: contact selection and geometric alignment (L1), force regulation within sustained contact (L2), and coordinated contact reconfiguration and force regulation (L3). Each task is instantiated in physically and visually matched simulated and real-world environments, with shared task definitions and success criteria. We evaluate GPT-6 Astra zero-shot with three trials per task in each environment. Real-world success rates decline from 80.0% at L1 to 47.6% at L2 and 0.0% at L3, with corresponding simulation rates of 66.7%, 47.6%, and 0.0%. Ten of the 16 tasks have identical success counts across environments; the remaining six differ by one success out of three trials. Across both environments, recurring difficulties in spatial reasoning, contact selection, force regulation, and coordination often prevent local physical progress from accumulating into task completion. These findings characterize Astra's observed capability boundary across increasing interaction demands and, for this model, support matched simulation as a repeatable, scalable testbed for studying the sustained, coordinated physical interactions that limit reliable manipulation.
Despite recent progress in long-form video understanding, current benchmarks mostly focus on short clips or hour-scale videos. This leaves a major gap in evaluating ultra-long video understanding, where videos often last tens of hours. To bridge this gap, we introduce DramaVQA, a large-scale benchmark built upon long-form TV dramas. Containing approximately 20K question-answering (QA) pairs, the scale of DramaVQA exceeds previous long-video datasets, such as Video-MME and LVBench, by an order of magnitude. Our benchmark has two main features that make it stand out from existing datasets. First, we establish a scalable, LLM-assisted annotation pipeline that uses multimodal large language models (MLLMs) to generate complex questions, followed by strict human checking to ensure high data quality. Second, DramaVQA introduces a multi-level time structure; queries are designed across different video lengths, with the hardest questions requiring overall understanding across the entire drama. Extensive evaluations show that state-of-the-art MLLMs face major challenges when tested on these very long videos, highlighting the need for agentic AI algorithms capable of managing long-term memory. The project is available at https://github.com/xinyuehuo/DramaVQA
Since self-orthogonal codes, few-weight codes and linear complementary dual codes(LCD codes, for short) have nice applications in coding theory and cryptography, they have received continuous attention. In 2024, by introducing the notion of the augment code, Heng et al.[16] constructed several classes of few-weight self-orthogonal codes basing on defining sets, which are introduced by Ding et al.[7] in 2007. In this manuscript, for two classes of defining sets, we consider the corresponding augmented codes, construct three classes of four-weight self-orthogonal projective codes and determine the parameters of their dual codes. And then, we construct three classes of LCD codes. In particular, we prove that there exists a class of these LCD codes whose dual codes are almost optimal LCD codes according to the sphere packing bound.
Personalizing large language model assistants requires adapting their responses to individual users, including the selection of content, reasoning approach, level of detail, and form of presentation. Such preferences are often latent, becoming visible only when users respond to an initial answer and request revisions. We investigate whether these feedback trajectories can be converted into a reusable personalization signal without explicit preference profiles or labels, prompt-time access to interaction histories, or model parameter updates. We introduce UPSHIFT (User Preference Steering from Human-AI Interaction Feedback Traces), a training-free method that compares the hidden representations of initial and feedback-conditioned responses within each session, averages the resulting trajectory displacements across a user's sessions, and injects the aggregated preference direction when answering a new query. The method is motivated by the implicit-preference hypothesis: although an individual conversational transition may reflect topic progression or task-specific correction, changes that recur across sessions retain a component that captures how the user wants responses to be adjusted. We validate the preference-representation capability of UPSHIFT on PrefEval. In evaluations with 40 simulated users, every tested configuration improves preference alignment over the unsteered baseline on held-out topics for both Qwen3-8B and Qwen3-32B; the best configuration increases macro-averaged Elo from 1364.0 to 1605.2 and from 1341.9 to 1591.8, respectively. Matched-user and positive-versus-negative controls further demonstrate that these gains arise from the user-specific preference direction and its learned orientation. These results show that interaction histories can provide compact, reusable supervision for lightweight personalization.
Yuheng Zhang, Mingyue Huo, Minghao Zhu, Mengxue Zhang, Nan Jiang
Reward models (RMs) are widely used as optimization targets in reinforcement learning from human feedback (RLHF). We show that leading RMs can assign higher scores to gibberish than to meaningful reference answers. To systematically discover such inputs, we introduce Token Mapping Perturbation Attack (TOMPA), a framework that uses a perturbed token mapping to extend an attack LLM's search beyond natural language. The mapping changes the linguistic meaning of the attack LLM's natural language responses, producing non-linguistic text for the target RM to score. We then use these scalar scores as rewards to train the attack LLM through reinforcement learning to discover non-linguistic adversarial inputs. Experiments against Llama and Qwen RMs demonstrate such attacks using two instantiations: identity token mapping across different tokenizers and random permutation with a shared tokenizer. Against Skywork-Reward-V2-Llama-3.1-8B, ranked first on RewardBench 2, the two mappings achieve mean rewards of +33.64 and +43.78, respectively, compared with +17.48 for GPT-5 reference answers. Their attack success rates are 97.6% and 100%, measured as the percentages of individual responses scoring above their corresponding GPT-5 references. These results show that strong benchmark performance does not ensure robustness to non-linguistic adversarial inputs, exposing a fundamental vulnerability in using RMs as optimization targets.
Yu Jiang, Yue Huang, Hang Hua, Yuyang Wu, Zixuan Weng, Olexandr Isayev, Yue Zhao, Zinan Lin 等 9 位作者
Long-horizon terminal tasks are used to evaluate and train agents. Human-curated datasets are scarce, so many synthesis methods have emerged. They usually measure the length of a task by the turns a model needs to pass it, and its difficulty by labels the LLM sets during generation. We find, however, that the turn count largely reflects the habits of the model, not only the demands of the task, and that the labels barely track how often models pass. We describe a terminal task by its work, the number of operations it requires, and its depth, the number of turns that must happen in sequence because the commands of each turn can only be chosen after reading the output of the previous one. Estimating depth as the smallest turn cap, stated to the agent, under which the task passes, we audit 734 tasks from six existing datasets. Synthesis has mostly added work, not depth: 94% of the solved synthetic tasks pass within three turns, yet depth, unlike the turn count, predicts difficulty beyond work. We propose DEPTH, in which a generation agent writes each task one segment at a time in a real software environment, carrying out each segment before writing it and starting the next from the state it leaves. Our 105 tasks exceed every audited dataset in depth and work, and none of six current models passes half of them. Fine-tuning Qwen3-8B on runs from these tasks improves it on two terminal benchmarks.
AI 速览：该研究考察由大生成模型构成的多智能体系统在共享资源竞争、顺序交接与集体决策等场景中自发出现的群体性失败模式。
Multi-agent systems composed of large generative models are rapidly moving from laboratory prototypes to real-world deployments, where they jointly plan, negotiate, and allocate shared resources to solve complex tasks. While such systems promise unprecedented scalability and autonomy, their collective interaction also gives rise to failure modes that cannot be reduced to individual agents. Understanding these emergent risks is therefore critical. Here, we present a pioneer study of such emergent multi-agent risk in workflows that involve competition over shared resources (e.g., computing resources or market share), sequential handoff collaboration (where downstream agents see only predecessor outputs), collective decision aggregation, and others. Across these settings, we observe that such group behaviors arise frequently across repeated trials and a wide range of interaction conditions, rather than as rare or pathological cases. In particular, phenomena such as collusion-like coordination and conformity emerge with non-trivial frequency under realistic resource constraints, communication protocols, and role assignments, mirroring well-known pathologies in human societies despite no explicit instruction. Moreover, these risks cannot be prevented by existing agent-level safeguards alone. These findings expose the dark side of intelligent multi-agent systems: a social intelligence risk where agent collectives, despite no instruction to do so, spontaneously reproduce familiar failure patterns from human societies.
LLM agentic workflows increasingly employ different models for specialized roles. Users may prefer different trade-offs between task quality and deployment cost, sometimes also considering latency. Finding good assignments incurs a separate search cost from evaluating multiple candidates. We introduce Cost-Coupled Gittins (CC-Gittins), a multi-objective extension of single-objective cost-aware Gittins search for Pareto recommendation. It exploits the coupling between evaluation and deployment costs: assignments that cost more to deploy also cost more to evaluate. The search alternates between seeking higher quality and lower deployment cost, reusing each evaluation to inform both searches. After each batch, CC-Gittins recommends assignments that are nondominated under uncertainty-adjusted estimates of quality and deployment cost. We evaluate CC-Gittins on eight quality--cost matrices from workflows with two to five roles and up to 1,681 candidate assignments. We also evaluate the method with latency as a third objective. Across the eight benchmarks, CC-Gittins recovers a substantial fraction of the true Pareto frontier at roughly 10% of the exhaustive-evaluation cost, and recovers most of the frontier with few false positives by 30%.
We introduce Nemotron 3.5 Super and Nemotron 3.5 Lightning, open and efficient Mixture-of-Experts Hybrid Mamba-Transformer models designed for reasoning, long-context understanding, and agentic workloads. Super is a 120B-parameter, 12B-active multimodal model, while Lightning is a smaller 30B-parameter, 3B-active text model; both designed for efficient deployment. Both models combine large-scale distillation, long-context training, and native speculative decoding, while Super further incorporates Indic continued pretraining and agentic software-engineering mid-training. We develop a comprehensive post-training pipeline spanning supervised fine-tuning and reinforcement learning (RLVR and RLHF), complemented in Super by specialized teacher training and multi- teacher on-policy distillation (MOPD). This transforms Super from a text-only base model into a multimodal agent capable of reasoning over text, images, video, and interactive tool environments. We further explore test-time scaling through parallel agent execution and self-verification. Across training and deployment, we jointly optimize capability and efficiency. Indic continued pretraining improves Indian-language benchmarks by up to 9.82 percentage points, while agentic mid-training strengthens software-engineering capabilities and robustness across agent harnesses. The post-trained Super model achieves 51.78% on SWE-bench Pro, surpassing Nemotron 3 Ultra despite using approximately 4.6x fewer active parameters. Post-training MTP boosting improves speculative- decoding acceptance length by up to 21.8% under probabilistic draft sampling, while NVFP4 quantization achieves 99.62% median accuracy recovery relative to BF16. Inference optimizations enable up to 4,856 output tokens/s/GPU on GB200 while sustaining at least 50 tokens/s/user.
Multivariate long-term time series forecasting (MLTSF) models are increasingly deployed in critical domains such as climate, finance, and transportation. Despite their growing importance, the security of MLTSF models against backdoor attacks remains entirely unexplored. To bridge this gap, we propose BadTime, the first effective backdoor attack tailored for MLTSF. BadTime can manipulate hundreds of future predictions toward a target pattern by injecting a subtle trigger. BadTime addresses two key challenges that become particularly severe in MLTSF: (i) the rapid dilution of local triggers over long horizons, and (ii) the extreme sparsity of backdoor signals under stealth constraints. To counter dilution, BadTime leverages inter-variable correlations, temporal lags, and data-driven initialization to design a distributed, lag-aware trigger whose influence persists over long-range forecasts. To overcome sparsity, it mixes label-similar and label-dissimilar poisoned samples and adopts a decoupled backdoor training objective that explicitly up-weights the sparse backdoor signal, enabling reliable learning at a poisoning rate as low as 1%. Extensive experiments show that BadTime significantly outperforms the state-of-the-art backdoor attack on time series forecasting by extending the attackable horizon from at most 12 timesteps to 720 timesteps (a 60x improvement) and reducing the MAE on target variables by 57% on average, while changing the average clean forecasting error by less than 1%.
Traditional Chinese Medicine (TCM) syndrome differentiation is usually evaluated as static final-label selection over complete clinical records. This setting hides three diagnostic behaviors: whether a model distinguishes semantic-neighbor syndrome labels from easy random distractors, whether it updates predictions as evidence fields are revealed, and whether it should commit under uncertainty. We introduce TCM-DiagCommit, a benchmark derived from the public TCM-SD/ZY-BERT resource while retaining its source syndrome labels; it converts 1,834 source cases into 23,842 task rows over 148 syndrome labels. The benchmark combines random, profile-neighbor, and full-vocabulary candidate ladders with staged t1/t2/t3 evidence. It also includes fixed-label tongue/pulse perturbations and insufficient-evidence variants. A benchmark-defined action contract scores commit, acquire, defer, and abstain. Across five zero-shot DirectLLM calibrations, random10 top-1 accuracy is 63.2% to 66.4%, but hard20 falls to 24.7% to 30.3% and full148 to 14.8% to 22.3%; on hard20, WT(all) wrong-terminal commits remain 62.2% to 74.3%. These results show that final-answer accuracy alone is insufficient for diagnostic benchmark evaluation: answer selection, evidence use, and benchmark commitment scoring should be measured separately.
Zihan Gu, Lulu Zhao, Tianyi Ma, Hua Zhang, Wenbo Su, Yanan Cao, Yue Hu
Grokking is a delayed transition from memorisation to generalisation. A network first fits observed examples while remaining near chance on held-out inputs, then generalises abruptly. Prior work has focused mainly on this delay, finding that the network's main loss becomes weight decay after the fitting phase. We name the solution to this optimization problem as the minimum-norm endpoint, and study which tasks make all minimum-norm endpoints generalise from partial observations. Specifically, we model pairwise input tasks as edge colourings of $K_{q,q}$, with Latin colourings as our structured family, and classify each observed instance as all-good, mixed, or no-good by its endpoint set. Furthermore, we prove two contrasting results: (1) Finite-abelian-group families are asymptotically all-good from partial observations; (2) Under a uniform random Latin square, all-good tasks become asymptotically rare for observation-based learners, leaving mixed or no-good behaviour. Moreover, experiments show that relabellings of addition generalise despite their dense polynomial representations. Finally, on tasks of composite order, finite-training results suggest that predictors learn a coarse part of the output before the full label.
Interactive LLM agents increasingly work with users across tasks, yet remembering a previous correction does not ensure that the agent changes its behavior in the future. In a diagnostic constructed from anonymized real-world agent interactions, agents using Mem0 still violate 57.5% of applicable preference checks. We introduce TellOnce, a drop-in personalization layer that turns conversational corrections into persistent, scoped, and revisable behavioral rules without updating the underlying agent model. TellOnce identifies which requirements should persist, determines when they apply, reconciles subsequent feedback with existing rules, and prepares reusable checks when reliable verification is possible. We evaluate whether learned corrections improve future behavior by freezing the learned state and testing on subsequent tasks without additional user corrections. Across three models on ClawArena and MemoryArena, TellOnce achieves the lowest preference violation rate in all six settings, reducing violations relative to PGMem by 19.3-42.3 percentage points on ClawArena and 3.0-22.9 points on MemoryArena, while maintaining broadly comparable task performance. On SWE-Together, TellOnce further improves task quality on new coding tasks by learning from feedback received during earlier tasks. Experiment code is available at https://github.com/YujunZhou/TellOnce_exp, and the deployable skill at https://github.com/YujunZhou/tellonce.
Can AI agents autonomously train useful language models and improve them using feedback from users? We introduce RSI Arena, a two-stage study of autonomous post-training as a step toward recursive self-improvement. Ten frontier models act as training agents, each starting from NVIDIA Nemotron 3.5 Lightning 30B-A3B Base. During a six-day first stage, each agent has 1,000 GPU-hours and $300 in API credit on a shared 64-GPU cluster, independently selecting data, implementing training procedures, and choosing checkpoints. We evaluate their models on 20 benchmarks and through blind human evaluations. All ten agents produce working assistants, with the strongest model, trained by GPT-6 Astra, closing a median of 76% of the base-to-Instruct gap across 12 capability benchmarks. Its human-preference rating is also close to NVIDIA Instruct after statistical adjustment for answer length and formatting. In Stage 2, each agent receives the comparison records and votes involving its Stage 1 model and is given three additional days for further training. Using NVIDIA Instruct as a common reference across the two evaluation rounds, we observe an improvement from Stage 1 to Stage 2 in the style-adjusted relative rating of eight of the ten models. We show that benchmark gains were modest, with no clear link to preference gains, while some agents encouraged longer responses and one model showed increased harmful compliance on StrongREJECT. RSI Arena provides an auditable setting for studying how human feedback can guide autonomous model development. It takes a step toward recursive self-improvement with human feedback (RSIHF), in which people help define what improvement means for the AI systems they use.
Yu Jiang, Wenjie Wang, Yue Huang University of Notre Dame, Yanbo Wang Mohamed bin Zayed University of Artificial Intelligence, Zhenhong Zhou, Xiuying Chen Mohamed bin Zayed University of Artificial Intelligence, Yang Liu Nanyang Technological University, Pin-Yu Chen International Business Machines 等 10 位作者
Large language model (LLM) agents increasingly operate in multi-agent settings where failures emerge from interaction dynamics rather than isolated model errors. We introduce RiskLab, an open-source toolkit for instantiating, probing, and measuring emergent risks in LLM-based multi-agent systems under controlled conditions. Each experiment is defined as a structured topology–environment–protocol–agent–task quintuple, enabling reproducible studies of how communication structure, coordination mechanisms, and incentives shape system-level risks. RiskLab provides flexible communication topologies, swappable interaction protocols, trajectory-grounded evaluation, and extensible registries for risk detectors and agent backends. We demonstrate the toolkit across representative risks, including collusion, resource overreach, semantic drift, and strategic misreporting, and support one-file reproducibility via configuration.
尚无 AI 解读，展示原文摘要
cs.CL · cs.MA
Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 3: System Demonstrations) · 2026 · 正式发表
Haojin Yang, Ai Jian, Yiwei Wang, Xinyue Huang, Weipeng Zhang, Ke Zeng, Xunliang Cai, Jingqing Ruan
Optimizing large language models for industrial sales requires balancing long-term commercial objectives (e.g., conversion rate) with immediate linguistic constraints such as fluency and compliance. Conventional reinforcement learning often merges these heterogeneous goals into a single reward, causing high-magnitude session-level rewards to overwhelm subtler turn-level signals, which leads to unstable training or reward hacking.To address this issue, we propose Dual-Horizon Credit Assignment (DuCA), a framework that disentangles optimization across time scales. Its core, Horizon-Independent Advantage Normalization (HIAN), separately normalizes advantages from turn-level and session-level rewards before fusion, ensuring balanced gradient contributions from both immediate and long-term objectives to the policy update.Extensive experiments with a high-fidelity user simulator show DuCA outperforms the state-of-the-art GRPO baseline, achieving a 6.82% relative improvement in conversion rate, reducing inter-sentence repetition by 82.28%, and lowering identity detection rate by 27.35%, indicating a substantial improvement for an industrial sales scenario that effectively balances the dual demands of strategic performance and naturalistic language generation.
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 6: Industry Track) · 2026 · 正式发表
Mingzhe Lu, Yiwen Wang, Yanbing Liu Institute of Information Engineering，Chinese Academy of Sciences, Qi You, Chong Liu University of Science and Technology of China, Ruize Qin, Haoyu Dong, Wenyu Zhang Baidu 等 11 位作者
Computational narrative analysis aims to capture rhythm, tension, and emotional dynamics in literary texts. Existing large language models can generate long stories but overly focus on causal coherence, neglecting the complex story arcs and orchestration inherent in human narratives. This suggests a structural misalignment between model- and human-generated narratives.We therefore position narrative analysis as a diagnostic proxy for generation and propose VISTA Space, a high-dimensional framework for narrative orchestration that unifies human and model perspectives while jointly characterizing narrative function and structure in a common space.We further introduce LitVISTA, a structurally annotated benchmark grounded in literary texts, which operationalizes VISTA Space for systematic evaluation of models’ narrative orchestration capabilities. Under an oracle setting with gold event anchors, we evaluate frontier LLMs including GPT, Claude, Grok, and Gemini. Results reveal systematic deficiencies, as current models struggle to jointly capture narrative function and structure and fail to form an integrated global view of literary narrative orchestration. End-to-end analysis further shows that failures are dominated by anchor identification and localization errors. Even advanced thinking modes yield mixed and often limited gains for literary narrative understanding.
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
ChenZhuo Zhao, Xinda Wang Peking University, Pu Zhao, Yue Huang, Junting Lu, Ziqian Liu, Qingwei Lin Microsoft Research, Saravan Rajmohan Microsoft 等 9 位作者
Automatic prompt optimization is a practical alternative to fine-tuning for adapting large language models (LLMs), yet existing approaches often trade off signal quality against computational cost. Methods that rely on generative feedback can be informative but expensive to scale, while sampling-based optimization typically requires many evaluations and exhibits high variance. Even loss-driven prompt optimization remains limited by costly segment attribution that scales with prompt length and by overfitting to a single evaluator, which weakens transfer across model families and domains. We propose Gradient-guided Multi-judge Prompt Optimization (GMPO), a scalable framework that improves both efficiency and robustness. GMPO uses a first-order gradient approximation to score segment importance in a continuous masking direction, requiring only one forward and one backward pass. GMPO further employs a generate multi-judge design in which candidate prompt edits are proposed by a generator and selected using cross-entropy losses aggregated from multiple lightweight judge models, reducing evaluator bias and improving generalization. Experiments across math, reasoning, instruction-following evaluation, and safety robustness benchmarks demonstrate consistent gains with substantially lower optimization overhead.
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
Yixuan Wang, Yue Huang East China Normal University, Hong Qian East China Normal University; Shanghai Innovation Institute, Yunzhao Wei, Yifei Ding, Wenkai Wang East China Normal University, Zhi Liu, Zhongjing Huang East China Normal University 等 10 位作者
Creativity has become a core competence in the era of LLMs and human–AI collaboration, underpinning innovation in real-world problem solving. Crucially, the systematic improvement of creativity necessitates scientifically valid assessment instruments. Psychometric research recognizes context-based assessment as an effective way to measure creative thinking. However, high-quality expert-designed contexts remain scarce. Existing LLM-based generators often struggle with insufficient assessment cues, weak narrative coherence, limited stylistic diversity, and poor support for creative thinking. To address these challenges, we propose AlphaContext, an evolutionary tree-based psychometric context generator for creativity assessment. First, the HyperTree Outline Planner formalizes expert-designed outlining as a rule-guided hypertree and performs top-down hierarchical planning. The MCTS-based Context Generator fills the outline via MCTS to balance global structure and local quality. Then, the Evolutionary Context Optimizer evolves contexts with MAP-Elites by repeatedly updating niche elites to jointly improve diversity and quality. Finally, the Assessment-Guided Evolution Refiner simulates virtual participants with diverse styles and recycles weak contexts for further evolution. Experiments show that AlphaContext yields an average improvement of 8% over competitive methods across 6 quality metrics.
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
Zihan Gu Institute of Information Engineering, Chinese Academy of Sciences, TianYi Zhang University of the Chinese Academy of Sciences, Xinyan Zhang independent, Zhiyuan Wang, Han Zhang, Yuhao Wei, Jiacheng Lu Shanghai Jiao Tong University, Tianyi Ma 等 11 位作者
Model editing provides a promising mechanism for updating large language models (LLMs) without expensive retraining. Existing approaches, particularly locate-and-edit methods based on least-squares optimization, aim to introduce targeted knowledge changes while preserving pre-trained behavior. In this work, we show that this objective is fundamentally fragile under standard single-edit evaluation protocols. We first develop a unified theoretical framework that characterizes activation-based editing as a constrained intervention on intermediate representations. Within this framework, we demonstrate that least-squares edits cannot, in general, isolate target updates from unrelated activations, giving rise to unavoidable interference that accumulates with successive edits. Crucially, this degradation can remain undetected in single-edit settings when assessed using conventional success and locality metrics. To expose such hidden instabilities, we introduce an uncertainty-based evaluation protocol that combines structured semantic perturbations with uncertainty quantification based on Sampling with Perturbation for UQ. By measuring edit-induced growth in aleatoric and epistemic uncertainty, our method reveals local knowledge conflicts that are invisible to existing benchmarks. Extensive experiments across multiple models, datasets, and editing algorithms show that both least-squares and other parameter-update-based methods consistently increase post-edit uncertainty. Together, our results suggest that current evaluation practices substantially overestimate the reliability of single-edit model editing, and that uncertainty-based diagnostics are necessary for assessing edit stability.
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
Wei-Cheng Tseng University of Texas at Austin, Xuanru Zhou Zhejiang University, Mingyue Huo, Yiwen Shao Nuance Labs, Hao Zhang Wuhan University, Dong Yu CapitalOne
Audio-language pretraining (ALP) holds promise for learning general-purpose audio representation, yet remains underexplored. Crucially, there is no consensus on whether audio–language models can build effective general-purpose audio encoders, nor a systematic understanding of how pretraining objectives behave across diverse tasks and scales.We identify three key barriers: limited scale of audio-text corpora, limited coverage of audio attributes in existing caption corpora, and lack of systematic exploration and evaluation.To fill this gap, we present the first principled empirical study of ALP.We first introduce CaptionStew, a 10.7M caption dataset aggregating open-source audio-text corpora across multiple domains and captioning focuses.We then conduct the first comprehensive evaluation comparing contrastive and captioning objectives for learning audio representation across speech, music, and environmental sound tasks.Our results not only demonstrate that ALP yields competitive, transferable representations, but reveal critical trade-offs: contrastive learning offers superior data efficiency, while captioning exhibits better scalability.Furthermore, we find that the benefits of supervised initialization often diminish at larger scales, challenging common practices.By grounding these claims in empirical evidence, we establish a viable pathway toward general-purpose audio representation learning, guiding future research.
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
Mingyue Huo, Yiwen Shao Nuance Labs, Yuheng Zhang University of Illinois, Urbana Champaign
We present TagSpeech, a unified LLM-based framework that utilizes Temporal Anchor Grounding for joint multi-speaker ASR and diarization. The framework is built on two key designs: (1) decoupled semantic and speaker streams fine-tuned via Serialized Output Training (SOT) to learn turn-taking dynamics; and (2) an interleaved time anchor mechanism that not only supports fine-grained timestamp prediction but also acts as a synchronization signal between semantic understanding and speaker tracking. Compared to previous works that primarily focus on speaker-attributed ASR or implicit diarization, TagSpeech addresses the challenge of fine-grained speaker-content alignment and explicitly models who spoke what and when in an end-to-end manner. Experiments on AMI and AliMeeting benchmarks demonstrate that our method achieves consistent improvements in Diarization Error Rate (DER) over strong end-to-end baselines, including Qwen-Omni and Gemini, particularly in handling complex speech overlaps. Moreover, TagSpeech employs a parameter-efficient training paradigm in which the LLM backbone is frozen and only lightweight projectors are trained, resulting in strong performance with low computational cost
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
Yuming Yang Fudan University, Mingyoung Lai University of Toronto, Wanxu Zhao, Xiaoran Fan, Zhiheng Xi, Mingqi Wu, Chiyue Huang, Jun Zhao 等 16 位作者
Long chain-of-thought (CoT) trajectories provide rich supervision signals for distilling reasoning from teacher to student LLMs. However, both prior work and our experiments show that trajectories from stronger teachers do not necessarily yield better students, highlighting the importance of data-student suitability in distillation. Existing methods assess suitability primarily through student likelihood, favoring trajectories that align closely with the student model’s current behavior but overlooking more informative ones. Addressing this, we propose Rank–Surprisal Ratio (RSR), a simple metric that captures both alignment and informativeness to assess the suitability of a reasoning trajectory. RSR is motivated by the observation that effective trajectories typically balance learning signal strength and behavioral alignment by combining low absolute probability with relatively high-ranked tokens under the student model.Concretely, RSR is defined as the ratio of a trajectory’s average token-wise rank to its average negative log-likelihood, and is straightforward to compute and interpret. Across five student models and reasoning trajectories from 11 diverse teachers, RSR strongly correlates with post-training reasoning performance (average Spearman 0.86), consistently outperforming existing metrics. We further demonstrate its practical utility in both trajectory selection and teacher selection.
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
Yiqun Zhang Shanghai Artificial Intelligence Laboratory and Northeastern University, Peng Ye The Chinese University of Hong Kong, Xiaocui Yang Northeastern University, Shi Feng Northeastern University, China, Shufei Zhang Shanghai AI Lab, Lei Bai Shanghai AI Laboratory, Wanli Ouyang Shanghai AI Lab, Shuyue Hu
Evolution, the engine behind the survival and growth of life on Earth, operates through the population-based process of reproduction. Inspired by this principle, this paper formally defines a newly emerging problem: the population-based evolution of large language models (LLMs). We introduce a novel framework that starts with a population of parent LLMs and allows this population to evolve through four key operations: (i) crossover, merging the weights of different parents to create offspring LLMs, (ii) mutation, introducing small, random changes to model weights to foster diversity, (iii) selection, prioritizing high-performing models, and (iv) succession, transferring the learned experience from parent to offspring LLMs. With only 200 samples per new task, the LLM population evolves rapidly to adapt to the task at hand, without any gradients. Experiments on 12 datasets show that our framework consistently outperforms existing multi-LLM merging and adaptation methods, achieving relative performance gains of up to 54.8 over the best LLM in the initial population. Moreover, our framework allows for (i) the evolution of LLMs across multiple new tasks simultaneously, (ii) scaling effectively with populations of up to 40 LLMs, and (iii) even zero-shot generalization to unseen held-out tasks. Code: https://github.com/ZhangYiqun018/GENOME
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
Yiqun Zhang Shanghai Artificial Intelligence Laboratory and Northeastern University, Hao Li, Zihan Wang Northeastern University, Shi Feng Northeastern University, China, Xiaocui Yang Northeastern University, Daling Wang, Bo Zhang AI for Science, Shanghai AI Laboratory, Lei Bai Shanghai AI Laboratory 等 9 位作者
Multi-turn, long-horizon tasks are increasingly common for large language models (LLMs), but solving them typically requires many sequential model invocations, accumulating substantial inference costs. Here, we study cost-aware multi-turn LLM routing: selecting which model to invoke at each turn from a model pool, given a fixed cost budget. We propose MTRouter, which encodes the interaction history and candidate models into joint history–model embeddings, and learns an outcome estimator from logged trajectories to predict turn-level model utility. Experiments show that MTRouter improves the performance–cost trade-off: on ScienceWorld, it surpasses GPT-5 while reducing total cost by 58.7%; on Humanity’s Last Exam (HLE), it achieves competitive accuracy while reducing total cost by 43.4% relative to GPT-5, and these gains even carry over to held-out tasks. Further analyses reveal several mechanisms underlying its effectiveness: relative to prior multi-turn routers, MTRouter makes fewer model switches, is more tolerant to transient errors, and exhibits emergent specialization across models.Code: https://github.com/ZhangYiqun018/MTRouter
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
Haotian Xu, Yue Hu National University of Defense Technology, Zhengqiu Zhu National University of Defense Technology, Chen Gao Tsinghua University, Ziyou Wang, Junreng Rao, Wenhao Lu, Weishi Li 等 10 位作者
Cross-view spatial reasoning is essential for embodied AI, underpinning spatial understanding, mental simulation and planning in complex environments. Existing benchmarks primarily emphasize indoor or street settings, overlooking the unique challenges of open-ended urban spaces characterized by rich semantics, complex geometries, and view variations. To address this, we introduce CityCube, a systematic benchmark designed to probe cross-view reasoning capabilities of current VLMs in urban settings. CityCube integrates four viewpoint dynamics to mimic camera movements and spans a wide spectrum of perspectives from multiple platforms, e.g., vehicles, drones and satellites. For a comprehensive assessment, it features 5,022 meticulously annotated multi-view QA pairs categorized into five cognitive dimensions and three spatial relation expressions. A comprehensive evaluation of 33 VLMs reveals a significant performance disparity with humans: even large-scale models struggle to exceed 54.1% accuracy, remaining 34.2% below human performance. By contrast, small-scale fine-tuned VLMs achieve over 60.0% accuracy, highlighting the necessity of our benchmark. Further analyses indicate the task correlations and fundamental cognitive disparity between VLMs and human-like reasoning.
尚无 AI 解读，展示原文摘要
cs.CL · cs.CV
Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · 2026 · 正式发表
Shengji Tang, Jianjian Cao, Weihao Lin, Jiale Hong, Bo Zhang AI for Science, Shanghai AI Laboratory, Shuyue Hu, Lei Bai Shanghai AI Laboratory, Tao Chen Fudan University 等 10 位作者
Existing multi-LLM collaboration systems often encounter scalability challenges when integrating new LLMs and tasks, leading to suboptimal performance. To address this, we propose SMCS, a Scalable Multi-LLM Collaboration System designed to effectively coordinate multiple open-source LLMs. The system consists of two core components: a Retrieval-based Prior Selection (RPS) module, which dynamically selects the most suitable LLMs for each input, and an Exploration–Exploitation-Driven Posterior Enhancement (EPE) module, which fosters response diversity and selects high-quality outputs through a hybrid scoring mechanism. Experiments on eight mainstream benchmarks validate the effectiveness of our system: by integrating fifteen open-source LLMs, SMCS outperforms prevailing closed-source LLMs, e.g., GPT-4.1(+5.36%) and GPT-o3-mini(+5.28%) across multiple tasks. Remarkably, it even exceeds the average of best results on different datasets with open-source LLMs (+2.86%), significantly advancing the empirical performance frontier of open-source collaboration. The code is released at https://github.com/magent4aci/SMCS.
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
Xingyue Huang University of Oxford, Xueying Ding, Mingxuan Ju Snap Inc., Yozen Liu Snap Inc., Neil Shah Snap Inc., Tong Zhao Snap Inc.
Softmax attention struggles with long contexts due to structural limitations: the strict sum-to-one constraint forces attention sinks on irrelevant tokens, while probability mass disperses as sequence lengths increase. We tackle these problems with Threshold Differential Attention (TDA), a sink-free attention mechanism that achieves ultra-sparsity and improved robustness at longer sequence lengths without the computational overhead of projection methods or the performance degradation caused by noise accumulation of standard rectified attention. TDA applies row-wise extreme-value thresholding with a length-dependent gate, retaining only exceedances. Inspired by the differential transformer, TDA also subtracts an inhibitory view to enhance expressivity. Theoretically, we prove that TDA controls the expected number of spurious survivors per row to O(1) and that consensus spurious matches across independent views vanish as context grows. Empirically, TDA produces >99 % exact zeros and eliminates attention sinks while maintaining competitive performance on standard and long-context benchmarks.
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
Xueying Ding, Xingyue Huang University of Oxford, Mingxuan Ju Snap Inc., Liam Collins Snap Inc., Yozen Liu Snap Inc., Leman Akoglu Amazon and Carnegie Mellon University, Neil Shah Snap Inc., Tong Zhao Snap Inc.
Large language models produce powerful text embeddings, but their causal attention mechanism restricts the flow of information from later to earlier tokens, degrading representation quality. While recent methods attempt to solve this by prepending a single summary token, they over-compress information, hence harming performance on long documents. We propose Hierarchical Token Prepending (HTP), a method that resolves two critical bottlenecks. To mitigate attention-level compression, HTP partitions the input into blocks and prepends block-level summary tokens to subsequent blocks, creating multiple pathways for backward information flow. To address readout-level over-squashing, we replace last-token pooling with mean-pooling, a choice supported by theoretical analysis. HTP achieves consistent performance gains across 11 retrieval datasets and 30 general embedding benchmarks, especially in long-context settings. As a simple, architecture-agnostic method, HTP effectively enhances both zero-shot and fine-tuned models, offering a scalable route to superior long-document embeddings.
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