Artificial Intelligence
When Context Changes: Understanding Update Failures in LLMs
Junyu Guo, Yuchen Fang, Shangding Gu et al. (+3 more)
As preferences, goals, and facts change, LLM
agents must use the current state while earlier versions remain in context. Yet they can answer with an old value of the same variable, a failure that we call stale binding. To study when models use outdated information and why, we introduce Controlled In-Context
Memory (CICM), a
benchmark for tracking and using updated information in conversations and
agent logs. We observe that even frontier
reasoning models can fail to recover the current state. We find that in open-source models probes can still recover the updated value when the model answers with an old one, pointing to a failure to select information that remains available. Component tests in Qwen and Pythia identify a mechanism for this selection failure:
attention drift, where
attention favors old values over the current one when producing an answer. We study a one-layer
transformer to mathematically understand how this phenomenon happens: when
attention scores are similar, several old values can together receive more
attention than the current value. Guided by this explanation, we redirect
attention toward the current value without further training. When the current value is requested directly, adjusting this intervention for each input corrects most old-value errors across various model families while preserving nearly all initially correct answers. Reliable context management therefore requires more than remembering updated information: models must use it to guide their answers.
Read the full abstract ▾
All authors Junyu Guo, Yuchen Fang, Shangding Gu, Costas Spanos, James Demmel, Javad Lavaei
Artificial Intelligence · Cross-listed from cs.LG
GeoNest: Learning to Select Failure-Aware Neighborhoods for the Irregular Knapsack Problem in a Circular Container
Zhongman Du, Huiming Zhang, Linlin Yang et al. (+2 more)
The two-dimensional irregular knapsack problem in a fixed circular container is an important combinatorial optimization problem for maximizing material utilization in manufacturing. Conventional
geometric packing solvers can produce tightly packed layouts, yet they often partition the residual space into isolated small pockets that cannot fit valuable unplaced polygons. To overcome this late-stage packing bottleneck, we propose a failure-aware large neighborhood search framework named GeoNest, driven by a graph policy trained via reinforcement learning. Specifically, we first construct neighborhoods by pairing failed target polygons with residual pockets. We then use explanatory poses to identify the placed polygons that block candidate insertions. These diagnosed blocking relations define bounded, fixed-item repair subproblems for the underlying
geometric solver. Finally, the graph policy selects the most promising subproblem for execution. For evaluation, we introduce CircleNest-Bench, a
benchmark comprising 2,391 load-controlled instances from four contour sources, including a held-out industrial CAD source. Experimental results demonstrate that, under the same total time budget, GeoNest improves mean utilization over a state-of-the-art standalone packing solver by about 0.9% on average across the three main test sets and by about 0.6% on the held-out industrial set.
Read the full abstract ▾
All authors Zhongman Du, Huiming Zhang, Linlin Yang, Sheng Xu, Baochang Zhang
Artificial Intelligence · Cross-listed from cs.RO
Efficient Multi-Modal Planning with Reward-Guided Preference Optimization for Autonomous Driving
Chenglin Chen, Lujia Wang, Xinhu Zheng et al. (+2 more)
Safe and efficient trajectory planning is essential in autonomous driving. However, existing end-to-end approaches often fall short in both computational efficiency and safety guarantees. Methods based on imitation learning suffer from causal confusion, while rule-based scoring approaches often incur heavy computational overhead and suffer from objective misalignment. Additionally, preference-based methods rely on strict pairwise annotations, limiting data utilization. To overcome these limitations, we propose EMPlan, an efficient multi-modal trajectory planning method powered by reward-guided fine-tuning. We design a hybrid architecture that combines sparse anchors with an offset refinement module for efficient multi-modal trajectory prediction. Sparse anchors provide coarse trajectory candidates with low latency, which are subsequently refined by the offset module for higher prediction accuracy. To enhance safety without incurring additional inference costs, we adopt a two-stage training paradigm consisting of pretraining and reward-guided fine-tuning. During fine-tuning, we leverage rule-based reward signals and unpaired preference supervision to refine the pretrained policy toward safer trajectory selection. We evaluate EMPlan on the non-reactive NAVSIM
benchmark, where it strikes a favorable balance between planning accuracy and efficiency, demonstrating superior performance under real-time constraints.
Read the full abstract ▾
All authors Chenglin Chen, Lujia Wang, Xinhu Zheng, Jun Ma, Haoang Li
Artificial Intelligence · Cross-listed from cs.LG
Optimal Design for Active Preference Learning with Biased LLM Judges
Zhongman Du, Huiming Zhang, Haodong Zhu et al. (+1 more)
Learning from human preferences is central to large
language model (LLM)
alignment, but human preference annotation is costly. Active preference learning reduces this cost by selecting informative comparisons, and LLM judges can provide additional scalable feedback. However, the preferences of the judges may deviate from those of the target human population. Even after calibration on trusted reference data, active acquisition can shift the comparison distribution and expose residual judge bias. We therefore incorporate judge deviations into the acquisition design rather than relying on a separate calibration stage. Under joint estimation, comparisons that appear highly informative about the reward may also reflect judge bias and therefore provide less information about human preferences. To address this issue, we propose Nuisance-Adjusted Optimal Design (NAOD), a comparison-selection strategy that prioritizes policy-relevant target information after nuisance adjustment and uses the Frank-Wolfe
algorithm for optimization. Theoretically, we establish a sharp conditional local asymptotic minimax lower bound on policy risk and construct an
estimator that attains it. We further characterize the finite-sample cost of learning the nuisance representation and show that representation error can reverse an oracle design advantage. Finally, we validate these predictions experimentally and evaluate NAOD on Chatbot Arena data across 17 judges, 15 budget configurations, and 15 random cluster-level splits. NAOD reduces the mean regret of proxy policy by 29.1% relative to a matched target-information design, outperforms existing methods, and improves human-preference prediction on held-out data.
Read the full abstract ▾
All authors Zhongman Du, Huiming Zhang, Haodong Zhu, Baochang Zhang
Artificial Intelligence · Cross-listed from cs.RO
Online Evolution Strategy for Flow-Matching VLA Policies via Self-Supervised Trajectory Distribution Optimization
Gongxin Yao, Yongsheng Zhao, Jiayin Deng et al. (+4 more)
Vision-
Language-Action (VLA) models based on generative frameworks, such as Flow Matching, have recently achieved impressive performance in robotic manipulation. Unlike deterministic policies, Flow Matching enables VLA models to learn conditional action trajectory distributions, where latent noise vectors induce different actions under the same task scenario. However, we observe that these distributions are often ill-formed, with successful and failed behaviors coexisting while considerable probability mass remains in unfavorable regions. To this end, we propose Online-ES, an online adaptation framework for Flow Matching VLAs based on
Evolution Strategy (ES), which refines the learned action trajectory distribution through interaction feedback. Instead of
pruning the latent noise space, our method performs evolutionary exploration directly in the action trajectory space, where diverse trajectories generated by Flow Matching provide candidate solutions for adaptation. By perturbing sampled trajectories and evaluating their execution outcomes, we derive a self-supervised MSE objective that transfers the
evolution direction from trajectory space into model parameter space. Mathematically, we prove that the proposed objective provides an unbiased
estimator of the optimal
evolution direction. Moreover, we also incorporate failure experiences as negative feedback to regularize the
evolution direction, steering the policy away from previously explored failure regions. Experiments in both simulation and real-world environments demonstrate that Online-ES achieves policy improvement comparable to reinforcement fine-tuning, without learning a value model or computing advantages.
Read the full abstract ▾
All authors Gongxin Yao, Yongsheng Zhao, Jiayin Deng, Deng Liang, Han Gao, Lei Zhao, Baoping Cheng
Artificial Intelligence · Cross-listed from cs.LG
Mitigating the Length-Scaling Tax with Online Distillation
Xu Wan, Wenyue Xu, Shengjie Zhao et al. (+1 more)
Length scaling during reinforcement-learning (RL) post-training is often viewed as a sign of improved
reasoning ability, especially on difficult problems, but may also make responses to already-solved problems unnecessarily verbose. We quantify this side effect as the length-scaling tax (LST): excess response length on already-solved queries without a commensurate accuracy gain. To mitigate LST, we propose Length Self-
Distillation (LSD), which routes solved
prompts to on-policy
distillation and retains the original RL objective for unsolved
prompts. LSD uses an exponential moving average of the online policy as its teacher, requiring no external model. We find that LSD achieves comparable or better performance than RL across multiple variants, while substantially curbing response-length growth on easy queries. LSD reduces LST from 19.0% to -3.7% on single-turn
reasoning and from 31.4% to 13.7% on multi-turn agentic tasks, demonstrating that LSD effectively preserves concise response patterns on easy queries while supporting efficient exploration on difficult queries during RL post-training.
Read the full abstract ▾
All authors Xu Wan, Wenyue Xu, Shengjie Zhao, Mingyang Sun
Artificial Intelligence · Cross-listed from cs.CL
Where MLLMs Fail and Why: Causal Task Decomposition for Capability Failure Diagnosis
Xia Hu, Brian Potetz, Chun-Ta Lu et al. (+6 more)
End-to-end accuracy on compositional tasks records how often MLLMs fail, but cannot distinguish whether a failure reflects an intrinsic deficit in the targeted capability or a cascading error from an upstream prerequisite. We propose a causal decomposition framework that isolates these two failure modes through controlled interventions on the prerequisite dependencies of each task. Our capability metrics (NC, IC, RC) score each task under unassisted, correct, or incorrect prerequisites to diagnose where failures arise; contribution metrics (N-Score, S-Score), adapted from probabilities of causation, quantify each prerequisite's necessity and sufficiency to determine why. We instantiate the framework in CADET, a diagnostic
benchmark of 10 composite tasks decomposed into 46 unit tasks with over 33,000 human-annotated questions spanning perception, spatial, temporal, and cognitive categories. Diagnosing frontier MLLMs with our framework uncovers systematic patterns that end-to-end accuracy obscures. Capability-wise, supplying correct prerequisites eliminates 54% of errors on cognitive tasks, lifting them from weakest to above spatial and temporal. Prerequisite-wise, causal contributions are concentrated in a few critical prerequisites, and supplying the single most important one alone captures 84% of the gain from supplying all prerequisites.
Read the full abstract ▾
All authors Xia Hu, Brian Potetz, Chun-Ta Lu, Huanfen Yao, Leonidas Guibas, Zhicheng Wang, Howard Zhou, Pengfei Xing, Andrew Gallagher
Artificial Intelligence
OpenJev-RLCD: A Working RLCD Implementation
Zhimin Gao and Pichao Wang
Decision models such as Jev answer questions with probabilities, which are only useful if they are calibrated. Open-source reproductions rely on supervised fine-tuning plus temperature scaling, while reinforcement learning from verifiable rewards (RLVR) makes
reasoning models overconfident. We present a working implementation of reinforcement learning for calibrated decisions (RLCD) for
reasoning models: the model samples a rationale, and we score the answer distribution it commits to afterwards with a strictly proper scoring rule. A variance identity shows that scoring the mixture of several samples rewards disagreeing rationales, and that RLVR is exactly this mixture objective without its diversity term. Optimized naively, the per-rationale objective either switches
reasoning off or is drowned out by policy-gradient noise, which leads to a two-stage recipe: calibrate, then reinforce. With Qwen3-1.7B on two
reasoning tasks (3 seeds, paired tests), RLCD matches or beats SFT, RFT/
STaR and GRPO (each temperature-scaled) in accuracy and beats all of them in selective prediction; on GSM8K answer verification a single query decides \gvTwoCovFive% of the items at $\le$5% error, versus \gvGrpoCovFive% for GRPO. When uncertainty comes from annotator disagreement, RLCD provably cannot beat cross-
entropy. Code and results: https://github.com/ZimmyGao/openjev-rlcd.
Read the full abstract ▾
Artificial Intelligence · Cross-listed from cs.LG
Scoring Higher, Answering Worse: Mitigating Reward Hacking in Rubric-Based RL via Protocol-Level Rubrics
Maoqi Liu, Junwei He, Bowen Zhang et al. (+5 more)
Rubric-based reinforcement learning (Rubric-RL) trains
language models where no verifier exists. A judge checks each criterion of a rubric, and the verdicts are aggregated into a reward, most often by a weighted sum. We show that this additive aggregation is the weak point. Under a sum, criteria compensate for one another: a policy that misses the one decision that matters can buy the points back with advice nobody asked for. On clinical consultation, such a policy scores higher and answers worse. Rubric coverage rises while appropriateness on held-out physician criteria falls below the untrained model. The medical criteria are not to blame. Grouped so that they must hold together, the same criteria, unchanged to the word, recover a third of the loss; shorter answers recover almost none. We therefore propose Protocol-level Rubrics (ProRubric), which keeps what the criteria ask for and changes how they are aggregated. It groups a checklist into a few protocol-level dimensions. A dimension counts only when all of its criteria hold and its failure clause does not fire. The grouping is done once, offline, and leaves the optimizer unchanged. ProRubric raises appropriateness by 10.8 points without losing coverage and has the best seven-
benchmark average at both scales. Reward validity is set not only by what a rubric verifies, but by how it aggregates. Code is available at https://github.com/Estrellajer/ProRubric
Read the full abstract ▾
All authors Maoqi Liu, Junwei He, Bowen Zhang, Feiran Li, Wentao Ma, Rongyi Lin, Shuhan Zhong, Quan Fang
Artificial Intelligence · Cross-listed from cs.LG
scTrilemma: Balancing Identity, Invariance, and Fidelity in Single-Cell Representation Learning
Yunhak Oh, Yoonho Lee, Junseok Lee et al. (+9 more)
Single-
cell RNA-seq representation learning is fundamentally label-free:
cell identities, states, and contexts are not fixed training targets, so what constitutes signal or nuisance is analysis-dependent. A single representation must therefore preserve biological identity and state, remain robust to nuisance context, and retain the gene-level variation needed for expression analysis, three demands we call the representation trilemma. To tackle this problem, we introduce scTrilemma, a latent-bottleneck VAE that routes expression-derived variation to the
embedding, the decoder, or the prior rather than forcing all of it through one
embedding. It gates gene
tokens by expression, routes the
cell representation through the decoder, and conditions the prior on unlabeled pseudo-bulk context, under a single reconstruction objective and without target annotations or auxiliary representation losses. In release-based zero-shot evaluation on successive CZ CELLxGENE Census releases, scTrilemma leads all three demands at once and preserves biological-state, differential-expression, and pathway structure across multiple disease settings. Latent interventions further show that context can be removed at almost no cost to the other demands, leaving identity against fidelity as the remaining tension. Code is publicly available at https://github.com/yunhak0/scTrilemma.
Read the full abstract ▾
All authors Yunhak Oh, Yoonho Lee, Junseok Lee, Namkyeong Lee, Sang-Yeon Hwang, Yinhua Piao, Hyomin Kim, Seonghwan Kim, Jaechang Lim, Woo Youn Kim, Sungsoo Ahn, Chanyoung Park
Artificial Intelligence
Diversity Combining for Multi-Path LLM Reasoning
Guangsheng Yu, Litianyi Zhang, Qin Wang et al. (+5 more)
Multi-path
reasoning methods such as self-consistency (SC) sample $K$
reasoning paths and choose the most frequent answer. However, their gains quickly plateau as $K$ increases, and existing methods do not predict when this saturation will occur. We formalize multi-path LLM
reasoning as a diversity combining problem from wireless communications: each path is a noisy channel observation, and the pairwise correlation of path correctness caps the design-effect effective sample size of the vote at a finite ceiling. Generalized least squares (GLS) analysis shows that, under exchangeability, the optimal symmetric linear combiner of latent
embeddings is uniform, supporting majority vote as the natural default in standard SC while leaving room for weighting or
pruning under heterogeneous
prompt-template branches. Across 5 models and 12
benchmarks,
prompt-template diversity reduces path correlation in $55$ of $57$ valid
cells, with the strongest effect on open-ended QA. We derive an Adaptive-K rule that uses a four-path pilot to select $K^*$, retaining $96$–$103\%$ of MV@$K{=}32$ accuracy across Math, QA, and NLU.
Read the full abstract ▾
All authors Guangsheng Yu, Litianyi Zhang, Qin Wang, Xu Wang, Mingyuan Li, Shaoxiong Ji, Ren Ping Liu, Massimo Piccardi
Artificial Intelligence
More Choices, Fewer Decisions: Ordinal-Scale Bias in JEV-like Direct-Decision Models
Tianxiang Gao, Jinzhe Li, Zhiyuan Li et al. (+2 more)
Direct-decision models enable efficient classification and evaluation by selecting from predefined candidates, but their reliability depends on more than prediction accuracy. Across 40
datasets and nine models from the JEV, KEV, and StartLux Decision families, we identify a systematic tendency to underuse ordinal rating scales. We call this phenomenon ordinal scale-utilization bias: final decisions concentrate on fewer effective labels than the gold distribution. The bias is substantially weaker on tasks with unordered semantic categories. Evaluating the same items on finer ordinal scales strengthens compression despite controls for gold imbalance and candidate position. Randomizing candidate order reduces but does not eliminate compression, while widening ordinal gaps at fixed candidate count can improve accuracy without restoring utilization. For most models, concentrated decisions coexist with broad average candidate probabilities. Greater model size improves accuracy without consistently restoring scale utilization. Together, these findings characterize compression as strongly associated with ordinal structure and intensified by scale refinement, rather than fully explained by presentation position or local spacing. This bias can be effectively mitigated through targeted training, making scale utilization an actionable reliability target alongside accuracy. The code and data are available at https://anonymous.4open.science/r/jev_ordinal_scale_bia/.
Read the full abstract ▾
All authors Tianxiang Gao, Jinzhe Li, Zhiyuan Li, Yi Chang, Yuan Wu
Artificial Intelligence · Cross-listed from cs.IR
SkillSeek: Revisiting Agent Skill Retrieval at Marketplace Scale
Guanqun Yang, Wenlong Zhang, Tian Shi et al. (+1 more)
Anthropic's
Agent Skills package reusable procedural know-how for an LLM
agent into SKILL.md directories, and open-source aggregations have grown past 230,000 skills, making selection rather than authoring the bottleneck. The standing answer in the literature outsources selection to the
agent itself: an LLM-mediated
retrieval loop that rewrites queries and refines candidates inside the
agent's decision loop, paying LLM
tokens on every task. We present SkillSeek, an open-source two-stage skill retriever built from the standard IR recipe (a BGE-base bi-encoder feeding a small cross-encoder, exposed over MCP). Across a $4 \times 11$ grid of pool, backbone, and method on the 89-task SkillsBench
benchmark, SkillSeek reaches observed parity with the LLM-mediated loop of Liu et al. at essentially no extra cost: plain bm25 alone records a pass rate at or above their refined loop on three of four settings, and a small cross-encoder covers the remaining difference on the fourth. A first-stage recall ceiling explains the pattern, and total per-trial spend drops from USD 51.30 to USD 27.54 (within fifty cents of the no-skill baseline). Under the SkillsBench tasks and OpenHands harness we tested, this positions the standard IR recipe as a strong default for
agent-skill
retrieval, with LLM-mediated alternatives a natural fit for cases where deterministic methods fall short.
Read the full abstract ▾
All authors Guanqun Yang, Wenlong Zhang, Tian Shi, Ping Wang
Artificial Intelligence · Cross-listed from cs.CL
BARRAC: Adaptation of an English Aspect-based Sentiment Analysis Approach for Classification Tasks in Arabic Dialects
Ali Almutairi, Gelareh Mohammadi, Imran Razzak et al. (+1 more)
With the rapid growth of Arabic NLP, several models,
datasets and
benchmarks have been reported. This paper asks whether approaches developed for majority
languages like English can be adapted to Arabic tasks. We adapt an English aspect-based sentiment analysis framework to Arabic classification tasks and present the adaptation as BARRAC: Brainstorming
Alignment and Replaced Representation learning for ArabiC tasks. BARRAC replaces consumer-review attribute pools with Arabic linguistic devices and markers for dialectal sentiment, sarcasm, and dialect identification, and replaces noisy self-training with two-stage training. Evaluated on five Arabic dialect
datasets, BARRAC achieves a mean macro-F1 of 63.93%, outperforming the best few-label SOTA by 3%, and outperforming GPT-4o on four out of five tasks. Error analysis provides insights into remaining challenges. These results demonstrate that adapting task-specific approaches is a promising direction for Arabic NLP alongside adapting models,
datasets and
benchmarks.
Read the full abstract ▾
All authors Ali Almutairi, Gelareh Mohammadi, Imran Razzak, Aditya Joshi
Artificial Intelligence · Cross-listed from cs.CV
Future Video Generation Better Aligns with the Human Visual Cortex than Observed Video
Chang-Bae Bang, Hyungjin Chung and Byung-Hoon Kim
Studying the
alignment between the internal representations of
vision models and the responses of the visual cortex to the same observed visual stimuli has enabled us to better understand human visual processing. However, studies so far have largely overlooked the fact that the human
brain not only processes observed visual stimuli, but also predicts upcoming stimuli based on what has been observed. Accordingly, we hypothesize that internal representations for generating future
video frames are better aligned with the predictive nature of human visual processing than representations of the observed
video itself. To this end, we compare the
alignment between human
video-watching fMRI responses in the visual cortex and the internal representations from two types of
video diffusion models, an autoregressive (AR) model and its non-AR base model. We first conduct a within-model analysis of the AR
video diffusion model and show that the representations for future
video generation align better with the visual cortex than the representations of the observed
video. We then compare the internal representations of the AR model with those of its non-AR base model and again show that the representations for future
video generation align better with the visual cortex than the representations for observed
video reconstruction by the base model. Specifically, the
alignment of observed
video reconstruction is concentrated in lower-order visual cortex, whereas that of future
video generation is concentrated in higher-order visual cortex. Finally, we show in a human behavioral experiment that humans prefer
videos generated by amplifying the contributions of individual layers that align better with the visual cortex.
Read the full abstract ▾
Artificial Intelligence
Whose Voice Survives the Summary? A Voice-Retention Audit of LLM Employee Listening
Thilo Tamme, Anton Hantel and Bijan Khosrawi-Rad
Organizations increasingly route employee feedback to leaders through large
language model (LLM) summaries, an unaudited layer that silences already-spoken voice. We introduce a Voice Retention / Representation Ratio metric for representational bias in summarization and apply it to a bilingual (English/German) corpus of 2,586 free-text responses from a global professional service company. First, employees supply criticism more reliably than praise (withholding praise is 82 times more common). Second, across 45 leader-summaries the pipeline filters by popularity, not sentiment: criticism survives, yet a concern voiced once is dropped 86% of the time, with short and German-only content lost on the same axis (theme retention 0.14 vs 0.74; German directional). Controlling for frequency, sentiment has no independent effect; the harm is prevalence-driven, which sentiment-only audits miss. A targeted
prompt recovers only named themes. We contribute the metric, field evidence, and a disaggregated voice-retention card.
Read the full abstract ▾
Artificial Intelligence
Yuanhe Zhang, Ziwei Wang, Jie Ren et al. (+6 more)
Large
reasoning models (LRMs) improve performance on complex tasks through extended
reasoning, yet the same process can degenerate into redundant verification and persistent generation loops. Such uncontrolled
reasoning increases inference cost and creates risks of resource exhaustion and service degradation. However, existing mitigations largely truncate long outputs or react to surface repetition, and thus fail to distinguish normal thinking from uncontrolled
reasoning or explain how benign
reasoning degenerates into harmful behavior. In this paper, we operationalize LRM generation as four states and further introduce
Reasoning-state Analysis via Dynamic
Attention Responses (RADAR), which identifies the current
reasoning state in real time and characterizes how effective reflection can develop into uncontrolled generation. Guided by RADAR's analysis, we further realign abnormal
attention distributions toward patterns observed in normal requests and examine how this correction affects excessive reflection and persistent looping. Temporal analyses show that uncontrolled
reasoning is characterized by
attention distributions that deviate from normal generation, with abnormal trends becoming detectable before repetition begins. Correcting these deviations through
Attention Realignment consistently reduces looping while largely preserving benign performance. Together, RADAR provide a mechanistic account of how
reasoning becomes uncontrolled, offering actionable guidance for identifying critical failure stages and designing targeted runtime interventions.
Read the full abstract ▾
All authors Yuanhe Zhang, Ziwei Wang, Jie Ren, Haoran Gao, Zhenhong Zhou, Fanyu Meng, Cong Wu, Li Sun, Sen Su
Artificial Intelligence · Cross-listed from cs.CV
CRAFT: Causal Responsibility and Failure Tracing in Medical Vision Language Models
Chunzheng Zhu, Jiaqi Zeng, Hongbo Zhao et al. (+3 more)
As
vision language models are increasingly deployed in clinical diagnosis, under standing how they internally resolve competing visual and textual signals becomes a safety imperative. Existing mechanistic analyses remain confined to unimodal text and offer no explanation for why a single misleading sentence can override a correct image based diagnosis, or why a model commits to a confident answer despite insufficient visual evidence. We find that these two safety risks, arbitra tion failure where textual context overrides visual grounding and brake failure where the model commits without adequate evidence, are mediated by spatially disjoint
attention head populations: arbitration heads form a mid-to-deep wideband reflecting cross-layer evidence competition, while brake heads concentrate in a narrow middle-to-late layer band that regulates evidence sufficiency and abstention behavior. To ground these observations in causal circuitry, we introduce CRAFT, which localizes each failure mode to a minimal causal head set via dual criteria and verifies necessity and sufficiency through temporal probes and Tuned Lens trajectory analysis. Excising arbitration heads sharply reduces conflict following with negligible degradation on clean inputs, while excising brake heads restores ap propriate abstention under degraded visual evidence. The two interventions target spatially disjoint head sets and produce distinct corrective effects, underscoring the mechanistic separability of the failure modes. Experiments across multiple medical VQA
benchmarks and VLM architectures validate both the localization and inter ventions, demonstrating that the identified heads causally drive each failure mode and that targeted modulation generalises without retraining. The code is available at https://github.com/zhcz328/CRAFT.
Read the full abstract ▾
All authors Chunzheng Zhu, Jiaqi Zeng, Hongbo Zhao, Yihang Chen, Yijun Wang, Jianxin Lin
Artificial Intelligence · Cross-listed from cs.CL
StateTree: Enhancing Long-Term Dialogue Reasoning via Reinforcement Learning
Naen Xu, Wanqing Cui, Yibo Hu et al. (+5 more)
Large
language models deployed as personalized assistants must reason over long, evolving interaction histories. However, in long-term dialogue
reasoning, relevant evidence is scattered across sessions, preferences may be revised over time, and standard long-context training fails to address these challenges under data scarcity and prohibitive computational costs. We propose StateTree, a data-driven RL method that constructs a challenging auxiliary task from scarce dialogues with verifiable ground truth. StateTree augments multi-session dialogues with a tree-structured path-tracing task: key-value records are embedded across sessions to form a binary tree. Solving the task requires the model to traverse from root to leaf by retrieving records across sessions and comparing timestamps to resolve branches, then recover the hidden target question among distractor leaves. We apply curriculum RL training progressively increasing tree depth and introduce a compositional variant whose edges carry step-level
reasoning fragments, training the model to compose partial cues into coherent queries. Trained on 10K-
token contexts, StateTree generalizes to 128K
tokens without full-length RL costs and exhibits capabilities including cross-session
retrieval, temporal
reasoning, knowledge update, and compositional multi-hop
reasoning. StateTree outperforms both SFT and RL-based baselines while preserving short-context general
reasoning. StateTree-7B achieves gains up to +23.60% on LongMemEval (128k), and StateTree-14B reaches 59.00% accuracy on LongMemEval, surpassing QwenLong-L1-32B (45.20%).
Read the full abstract ▾
All authors Naen Xu, Wanqing Cui, Yibo Hu, Shixin Hong, Hengyu An, Meiguang Jin, Junfeng Ma, Tianyu Du
Artificial Intelligence · Cross-listed from cs.LG
Explicit Trajectory Diversity for RL-Based Post-Training of LLM Agents
Huaiyu Fu, Heng Cao, Hao Wang et al. (+2 more)
LLM
agents often admit multiple high-quality solutions to the same task, differing in
reasoning structure, tool-use pattern, or interaction trajectory. Yet existing notions of diversity in LLM post-training are mostly implicit, arising from general stochasticity and regularization mechanisms rather than explicitly targeting task-relevant behavioral variation. While such implicit diversity can be useful, it does not directly specify which forms of behavioral variation should be encouraged for a given task. In this work, we study explicit trajectory diversity in RL-based post-training for LLMs. Our key idea is to define diversity through user-specified, task-specific trajectory descriptors, which map each sampled trajectory to an interpretable behavioral representation, and then measure diversity as a set-level functional over the resulting descriptor
matrix. Building on this formulation, we introduce Trajectory-guided Joint Policy Optimization(TJPO), a single-policy framework that optimizes explicit diversity over sampled trajectory groups, avoiding the need for population-based policy training, and instantiate it within group-based policy optimization through trajectory-level learning signals. This design makes the diversity objective both interpretable and controllable. Experiments on Sokoban and ALFWorld show that TJPO improves task-specific trajectory diversity while maintaining competitive task performance. Descriptor and trajectory analyses show that the learned variation follows the specified behavioral dimensions and includes distinct successful strategies. Extra experiment results suggest that explicitly shaping trajectory diversity can help LLM
agents satisfy user requirements and remain effective when task conditions change.
Read the full abstract ▾
All authors Huaiyu Fu, Heng Cao, Hao Wang, Jian Ya, Tao Chen
Artificial Intelligence · Cross-listed from cs.CL
Uncovering Uncontrolled Repetition through Residual Stream Dynamics
Yuanhe Zhang, Xinyao Zhou, Haoran Gao et al. (+5 more)
Uncontrolled repetition can prolong autoregressive generation in large
language models (LLMs) and enable resource consumption attacks. Prior analyses of repetitive generation have identified strongly activated features in intermediate and late layers. However, how uncontrolled repetition activity emerges and develops before becoming prominent in these layers remains insufficiently understood. In this paper, we investigate this question primarily in large
vision-
language models (LVLMs), which support a richer set of uncontrolled repetitions through both visual and textual inputs. We propose Tokenwise Residual Comparison (TRC), a method that identifies and localizes anomalies associated with repetition from residual dynamics during generation. TRC compares
attention and multilayer perceptron writes to the residual stream across generated
tokens to identify patterns associated with repetition. It then selectively suppresses coordinates in the residual stream at the identified layer. Experiments show that TRC effectively mitigates uncontrolled repetition, reducing loop rates by 57% on average. Our analysis further shows that repetition semantics emerge in shallow layers and propagate through the residual stream, disrupting normal representations. TRC also generalizes to large
language models (LLMs) and large
reasoning models (LRMs), where it consistently captures analogous repetition dynamics and achieves effective mitigation. Our work broadens the study of repetitive generation from its prominent internal representations to earlier opportunities for intervention, providing insights for mitigating resource consumption attacks.
Read the full abstract ▾
All authors Yuanhe Zhang, Xinyao Zhou, Haoran Gao, Yuyao Zhang, Zhenhong Zhou, Fanyu Meng, Li Sun, Sen Su
Artificial Intelligence
GraphCert: Bootstrap Agentic Graph Reasoning with Certified Evidence Rubrics
Weiqi Jiang, Yuchen Ying, Rui Wang et al. (+5 more)
Graph
agents extend large
language models (LLMs) with the ability to actively explore and reason over knowledge graphs through multi-step interactions with graph tools. However, training capable graph
agents typically requires large collections of question-answer pairs and
reasoning trajectories, whose manual construction is costly and difficult to scale. Moreover, employing proprietary LLMs to generate such supervision further risks exposing sensitive graph data to external services. Therefore, we propose GraphCert to bootstrap agentic graph
reasoning with certified evidence rubrics during post-training. Specifically, the Bootstrapped Graph Quizzer guided by generation controls produces graph-grounded QA pairs and marks supporting evidence, which undergo execution certification and semantic curation. The accepted evidence is then canonicalized into certified evidence rubrics that later reward Graph Solver evidence
alignment alongside answer correctness during GRPO training. Experiments on five graph
reasoning domains in GRBENCH demonstrate that GraphCert consistently outperforms substantially larger LLM
agents and post-training method. Furthermore, our analysis demonstrates that the learned policy transfers robustly across heterogeneous graph domains, suggesting that GraphCert acquires reusable graph-
reasoning capabilities rather than domain-specific patterns. These results establish executable self-certification as an effective approach to self-training compact graph
reasoning agents. Our code will be made publicly available.
Read the full abstract ▾
All authors Weiqi Jiang, Yuchen Ying, Rui Wang, Kaixuan Chen, Bingde Hu, Shunyu Liu, Yu Wang, Tongya Zheng
Artificial Intelligence · Cross-listed from cs.LG
Evaluating Persistent Calibration under Evolving Model Knowledge
Victor Wang, Thomas Hofweber, Mohit Bansal et al. (+1 more)
As AI systems move from static repositories to
agents that are capable of continual adaptation and learning, maintaining their trustworthiness means equipping the models backing them with the ability to produce confidence estimates that dynamically reflect their changing skills and knowledge. We introduce the problem of persistent calibration, which requires a confidence
estimator to faithfully reflect the knowledge contained in a model as that knowledge changes, without recurring supervision. We operationalize this by examining persistent calibration across checkpoints of open models, asking whether confidence
estimators trained on earlier checkpoints can generalize to later ones. Specifically, we aim to shed light on whether confidence is dependent on knowledge, a question with implications for the reliability of confidence estimates. To measure this relationship, we define and evaluate calibration on knowledge contrast sets: subsets containing questions that one checkpoint answers correctly and another checkpoint answers incorrectly, reflecting a change in knowledge. We show that both inference-time and fine-tuning methods fall short on contrast-set calibration compared to oracle methods trained on future checkpoints, even for methods that are well-calibrated on the full
dataset. We provide evidence for the hypothesis that persistent calibration is challenging because there is a vast space of possible confidence functions that are well-calibrated on a given checkpoint, out of which only some rely on meta-knowledge features that would generalize to other checkpoints. Towards improving contrast-set calibration, we show that multi-checkpoint training helps, suggesting an avenue for identifying confidence features that remain robust across changing knowledge.
Read the full abstract ▾
All authors Victor Wang, Thomas Hofweber, Mohit Bansal, Elias Stengel-Eskin
Artificial Intelligence
Positive Ratings, Hidden Concerns: Employee Voice Disclosure in AI-Mediated Organizational Listening
Thilo Tamme, Michael Saatkamp, Alma Bonte et al. (+3 more)
Organizations started listening to employees through conversational AI
agents alongside structured surveys. Little is known about what these channels change in what employees say when disclosure carries hierarchical risk. We report a field study inside a global management consulting firm whose process pairs a pre-survey with an adaptive AI voice interview on the same themes within one session. Across 44 first-session interviews (132 matched theme observations), 20-41% of sessions showed a favorable rating co-occurring with a substantive concern voiced later, depending on the favorability threshold. The Gioia analysis drew on 158 protective quotes from 65 eligible sessions. Disclosure rarely arrived unguarded: employees softened concerns, deflected accountability, and bounded how far they went, and this protective work tracked the perceived legitimacy of the listening structure. We develop a grounded model of bounded disclosure and derive four propositions for voice, channel and listening research. Silence, we argue, can persist inside expression.
Read the full abstract ▾
All authors Thilo Tamme, Michael Saatkamp, Alma Bonte, Daniel Weiss, Anton Hantel, Andrej Levin
Artificial Intelligence
Persona and Persuasive Framing in AI Voice Agents: A $2\times2$ Field Experiment with Children
Thilo Tamme, David Steck and Anton Hantel
Conversational
agents increasingly interact with children, yet evidence on how their design shapes children's susceptibility to persuasion comes almost entirely from the lab. We report a $2\times2$ randomized field experiment embedded in a public German Santa Claus telephone hotline. Children's calls were randomly routed to one of four LLM voice
agents varying persona (Santa, high authority, vs. Helper, low authority) and framing (persuasive nudges toward prosocial wishes vs. neutral). Of 1,072 logged calls, 89 conversations (median age 6) met inclusion criteria. Persuasive framing raised the probability of a prosocial wish from 11.6% to 45.7%, robust to controls. Persona authority showed a near-zero effect: Santa did not outperform the Helper. Persona instead shaped engagement; children hung up on the Helper far more often within the first minute (65% vs. 39%). Where context already lends an
agent legitimacy, how it speaks shapes children's compliance more than who it claims to be.
Read the full abstract ▾
Artificial Intelligence · Cross-listed from cs.LG
ChartDensity-Bench: Benchmarking MLLMs for Numerical Data Reconstruction under Visual Density
Xinhe Wu and Yadong Jin
Multimodal large
language models (MLLMs) offer a promising approach for recovering numerical data from scientific charts, but their ability to reconstruct chart data from visually dense figures remains poorly understood. Existing chart understanding
benchmarks primarily evaluate question answering or chart-level
reasoning and provide limited support for evaluating structured numerical reconstruction from scientific figures. We introduce ChartDensity-Bench, a
benchmark for evaluating MLLMs on structured numerical data reconstruction from compound chart figures under controlled visual density. Built from charts paired with source-level ground-truth data, ChartDensity-Bench systematically varies the number of simultaneously presented charts ($k\in{1,3,6,9}$), enabling controlled evaluation of density-induced degradation. We further propose a multi-dimensional evaluation framework covering structural reliability, reconstruction completeness, parseability, and numerical fidelity. Experiments on five recent MLLMs show that numerical reconstruction generally degrades as visual density increases, while the magnitude of degradation varies substantially across models. Chart-level paired comparisons further show that the same source chart can incur higher reconstruction error when embedded in denser visual contexts. These findings highlight visual density as an important and previously underexplored factor in MLLM chart data reconstruction and provide a systematic
benchmark for evaluating model robustness in this setting.
Read the full abstract ▾
Artificial Intelligence · Cross-listed from cs.SD
RAST: Resolution-Aware Privileged Structure Transfer for Low-Resolution Audio Activity Recognition
Ji Hwan Park, Gautham Krishna Gudur, Yufei Shen et al. (+2 more)
Audio is increasingly used for human activity recognition (HAR) because it captures object interactions, environmental events, and contextual cues in everyday environments. High-resolution (HR) audio provides rich acoustic information for model development but incurs substantial energy and storage costs and may expose sensitive
speech content. Low-resolution (LR) audio offers a more
privacy-preserving and resource-efficient alternative for deployment, but reduced sampling rates can remove acoustic cues essential for activity recognition, leading to significant performance degradation. We formulate this training-deployment mismatch as sensor-resolution privileged learning, in which HR audio is available during training, while inference relies exclusively on LR audio. We propose RAST, a resolution-aware transfer framework that compresses HR teacher representations by preserving
token-level information and neighborhood structure before performing localized HR-LR
alignment. Experiments on the SAMoSA and AudioIMU
datasets show that RAST consistently outperforms LR-only training and direct teacher-transfer baselines, improving LR-only recognition by up to approximately 7.8% while requiring only LR audio at inference.
Read the full abstract ▾
All authors Ji Hwan Park, Gautham Krishna Gudur, Yufei Shen, Dawei Liang, Edison Thomaz
Artificial Intelligence
Action Conditioned Bisimulation For GUI Agent Memory
Hongbo Zhang, Liuyang Song, Quanquan Li et al. (+3 more)
An
agent that remembers what it did on a web page must decide when two pages count as the same.
Memories built on observation similarity merge pages that look alike but behave differently, and GUIs are full of such pages: two tabs of one widget or two rows of one menu answer the same click differently. We define the merge rule as an action-conditioned bisimulation over the empirical predictive state graph a frozen
agent fills as it acts. Two states merge only when their shared actions lead to agreeing outcomes and successor blocks under an affordance label. Observation similarity never enters the rule, and nothing is trained. It replaces the merge rule of an existing outcome-value
memory, so a closed-loop comparison isolates it. On MiniWoB++ it raises success rate over a memoryless
agent, while a control taking identical exploratory detours, the prior successor-representation merge, and the same criterion without action conditioning change nothing.
Read the full abstract ▾
All authors Hongbo Zhang, Liuyang Song, Quanquan Li, Daqian Yang, Yan Wen, Zhengtao Yao
Artificial Intelligence · Cross-listed from cs.LG
Distilling Diffusion Score Discrepancy for Efficient Training Data Attribution
Shixuan Liu, Joan Serrà, Kin Wai Cheuk et al. (+6 more)
Training data attribution for
diffusion models aims to identify the training samples that influence a generated instance, but existing methods either require costly per-sample gradient computation or query-specific model optimization. Moreover, most methods attribute changes in a proxy loss rather than changes in the actual model's generative behavior. We address these limitations by formulating attribution directly with a local score discrepancy measure, which applies to any
diffusion variant (including DDPM, EDM, and flow matching), and by showing that such measure can be estimated without retraining, as a preconditioned gradient similarity. We instantiate this
estimator as Training-data Influence via score Discrepancy (TID), which uses Kronecker-factored
curvature to avoid random projections and per-sample gradient storage. We then distill TID into TIDE, a forward-only student trained online to reproduce the teacher's rankings from the
diffusion model's internal activations. Under counterfactual evaluation on CIFAR-10, ArtBench-10, and MS-COCO, TID matches or outperforms state-of-the-art approaches, while TIDE retains most of TID's accuracy at four to five orders of magnitude lower per-query cost, attributing generated samples in milliseconds and faster than the generation itself.
Read the full abstract ▾
All authors Shixuan Liu, Joan Serrà, Kin Wai Cheuk, Jinju Kim, Woosung Choi, Yukara Ikemiya, Wei-Hsiang Liao, Jiaqi W. Ma, Yuki Mitsufuji
Artificial Intelligence · Cross-listed from cs.LG
Learning Under Forgetting: Statistical Support-Selective Retention in Stochastic Training Dynamics
Fujie Gao, Zuyue Zhang and Gang Sun
Prior work has shown that neural networks exhibit implicit biases toward low-complexity structure (e.g., spectral bias), memorization dynamics, and compression-like effects during training, but a unified dynamical account of selective retention remains incomplete. We propose Repeated Reinforcement with Persistent Forgetting (RPF) dynamics, a minimal framework in which repeated exposure reinforces patterns and structures that recur in the data, while persistent forgetting attenuates learned information. This view treats forgetting not merely as a failure mode, but as a selection mechanism. We build the theory in three successive layers. First, in an independent-feature model, we derive an exposure-selective survival law and a support-dependent retention boundary characterizing which patterns persist under forgetting. Second, in a shared-parameter model, we show that forgetting induces spectral filtering over covariance modes, preserving strongly supported shared components while suppressing weak ones. Third, under small-step and norm/coding approximations, we show how RPF dynamics induce an implicit trade-off between data fitting and the cost of stored information, yielding Minimum Description Length (MDL)-like compression. Controlled experiments provide evidence for this reinforcement–forgetting selection mechanism in scalar
memories and a nonlinear shared network. Joint reinforcement and attenuation interventions shift conditional retention, while matched exposure counts reveal forgetting-dependent effects of reinforcement timing and changes in the composition of the retained set. Together, these results show that repeated reinforcement and persistent forgetting jointly provide a controllable source of inductive bias beyond neural architecture and scale.
Read the full abstract ▾