Artificial Intelligence
Unlocking Token Rewards via Training-Free Reward Attribution
Sitong Wu, Haoru Tan, Bin Xia et al. (+6 more)
In this paper, we propose an extremely efficient, training-free method to extract
token-level reward signals directly from an existing deep reward model. Our core idea is to attribute the overall process reward to individual
tokens by estimating each
token's influence. This influence is defined as the change in the final macroscopic reward (e.g., the process reward) when a
token is replaced with a semantically null
token. Naively calculating this influence is computationally infeasible, requiring N forward passes through the PRM for an N-
token sequence. We overcome this bottleneck by proposing a highly efficient gradient-based
estimator. Specifically, we use a first-order Taylor approximation, which simplifies the influence calculation to the inner product of the difference between the
token embedding and the null
token embedding, and the gradient of the reward with respect to the
token embedding. This requires only a single forward and backward pass. The resulting
token-level rewards enable standard RL
algorithms to perform precise credit assignment without requiring additional reward model training. Experiments on challenging
reasoning benchmarks demonstrate that our method substantially improves policy optimization efficiency and enhances the generalization of LLM
reasoning capabilities. Our P2T outperforms the outcome reward by +4.9% on MathVista for Qwen2.5-VL-7B-Instruct, and +11.5% on AIME24 for Qwen2.5-Math-7B. The code will be released via https://github.com/stonewst/P2T.
Read the full abstract ▾
All authors Sitong Wu, Haoru Tan, Bin Xia, Xichen Zhang, Jingyao Li, Shaofeng Zhang, Xiaojuan Qi, Bei Yu, Jiaya Jia
Artificial Intelligence · Cross-listed from cs.RO
UniW(A)M: A Unified Mixture-of-Experts Framework for World and World-Action Models
Jingbo Xu, Jianjie Fang, Zhuohang Li et al. (+9 more)
World models (WMs) and world-action models (WAMs) focus on future world prediction and executable action generation, respectively. Although these models learn shared physical dynamics and interaction semantics from embodied trajectories, they are typically trained in isolation, which restricts knowledge reuse between prediction and control. The central challenge is therefore to develop a unified learning framework that shares these dynamics while preserving the distinct modeling capabilities required by each paradigm. We propose UniW(A)M, a mixture-of-experts world-(action) model built upon
diffusion Transformers that unifies action-conditioned world modeling and diverse WAM variants. Shared experts capture common dynamics, while atomic capability experts are reused across interfaces so that each interface keeps its own conditional pathway. At deployment, a plug-in adaptive interface router (PAIR) learns from a small subset of target-task chunk scenarios to select a suitable WAM interface for each execution chunk, balancing task success and execution efficiency while keeping the backbone frozen. We evaluate UniW(A)M on LIBERO, RoboTwin 2.0, and three real-world manipulation tasks, where it achieves 88.3% average real-world success versus 76.7% for the strongest baseline. On WorldArena, it leads action-conditioned world modeling on all three tasks. AC-WM co-training contributes to action generation on every action interface and PAIR raises OOD success from 63.3% to 76.7%.
Read the full abstract ▾
All authors Jingbo Xu, Jianjie Fang, Zhuohang Li, Zihao Song, Zhaolu Wang, Xvyuan Liu, Xin Zhang, Haisheng Su, Yinzhou Tang, Chen Gao, Wei Wu, Yong Li
Artificial Intelligence · Cross-listed from cs.CV
GeoReader: Learning Latent Geometry for Spatial Reasoning in Vision-Language Models
Rong Li, Shijie Li, Jiangyong Huang et al. (+7 more)
Vision-
language models (VLMs) struggle with metric and multi-view relationships that remain implicit in RGB observations. Geometry foundation models recover this structure from RGB, but existing methods either fuse the estimated features with visual
tokens and learn to use them from question answering alone, or align VLM latents with geometry features. Capturing geometry, however, does not ensure its use: zeroing the aligned latents of a released feature-
alignment model leaves its accuracy unchanged. We introduce GeoReader, a framework that learns
geometric representations for the VLM that reads them. First, we train lightweight
Geometric Readers to produce continuous input
tokens, supervising them with geometry descriptions decoded by a frozen VLM, so that the Readers express geometry in a form the target model can already read. We then freeze the Readers to preserve this interface while fine-tuning the VLM on spatial QA, with geometry injection conditioned on the question. Geometry descriptions serve only as training targets: at inference, the VLM answers directly from the continuous
geometric tokens without generating a geometry chain of thought. With geometry predicted from RGB images, our approach reaches 94.3 on MindCube-Tiny with Qwen3-VL-4B and, with a multi-source QA mixture, an average of 72.5 across six additional spatial
benchmarks with Qwen3.5-9B. These results suggest that a useful
geometric representation is one the receiving model can use, not merely one from which geometry can be reconstructed.
Read the full abstract ▾
All authors Rong Li, Shijie Li, Jiangyong Huang, Lingdong Kong, Jiaming Zhou, Dongyue Lu, Zeying Gong, Baoxiong Jia, Junwei Liang, Siyuan Huang
Artificial Intelligence · Cross-listed from cs.CV
Just Respect the Semantics: Image-Token Compression via Pre-LLM Semantic Variable-Resolution Tokenization and In-LLM Relevance Pruning
Tianhao Fu, Austin Wang, Charles Chen et al. (+5 more)
Image-
token compression must reduce computation without hurting the semantics. We introduce Varest, guided by semantic fungibility:
tokens within a coherent semantic region can be summarized coarsely, while object boundaries and literal text require finer distinctions. Object and text localizers guide variable-resolution merging, preserving spatial coverage and finer detail at boundaries and text. Learned [REL] registers then prune visual
tokens after early decoder processing. Training across budgets enables one checkpoint per backbone. Across three backbones and 13
benchmarks, Varest outperforms 15 baselines in mean performance retention at every evaluated budget, retaining 86.10-92.33% of full-
token performance with only 5.56% of the original layer-averaged image-
token count.
Read the full abstract ▾
All authors Tianhao Fu, Austin Wang, Charles Chen, Roby Aldave-Garza, Victor Xie, Xiyue Jiao, Yucheng Chen, Martin Tin
Artificial Intelligence · Cross-listed from cs.CV
RadarGSOcc: Doppler-Aware 3D Semantic Occupancy Prediction Using 4D Radar Tensor with Geometric-Semantic Feature Extraction
Christofel Rio Goenawan, Dong-Hee Paek and Seung-Hyun Kong
3D semantic occupancy prediction is crucial for autonomous driving, but camera-based methods degrade in low-light conditions and
LiDAR-based methods are sensitive to adverse weather. 4D radar is more robust, yet its sparsity and the high cost of radar tensor processing limit its use. We propose Radar
Geometric-Semantic Occupancy Prediction (RadarGSOcc), an efficient framework that directly processes 4D radar tensors by separately extracting
geometric and semantic features, including Doppler information, with a deformable
attention-based
geometric refinement module as its key component. We also introduce the K-Radar 3DSemOcc
benchmark, the first large-scale
benchmark for 3D semantic occupancy prediction using 4D radar tensors, built on the K-Radar
dataset with refined
LiDAR-based annotations. RadarGSOcc predicts 15 classes within a 51.2m range and achieves 38.2% IoU, 7.6% mIoU, and 25.5% weighted mIoU, outperforming RadarOcc by 17% relative (25.5 vs. 21.9) in weighted mIoU while using fewer parameters and lower computation. It substantially improves truck prediction while trading off car accuracy, outperforms camera-based approaches, and narrows the gap with
LiDAR-based methods. To assess robustness without per-frame ground truth in adverse weather, we further propose a novel proxy metric, Environment-Based KL Divergence (EBKL), under which 4D radar proves a reliable and efficient modality for 3D semantic occupancy prediction. The
dataset and code will be publicly available.
Read the full abstract ▾
Artificial Intelligence
Knowing the User Is Not Knowing What We Established: Grounding Awareness in Personal Agents
Jiajun Chen, Xinyue Chen, Yicheng Jiang et al. (+1 more)
Personal-
agent memory has largely focused on learning what users know, prefer, and want. However, knowing the user does not necessarily mean knowing what the user and the
agent have established together. An assistant's proposal should not become an authorized instruction merely because it is remembered. We introduce Grounding Awareness, the ability to distinguish accepted, unacknowledged, revised, and scoped arrangements when selecting actions. We develop a controlled
benchmark to test whether
agents can tell the difference between proposals users have accepted and those they have not, and whether they apply accepted arrangements only where appropriate. Across multiple
agents and
memory systems, we find that strong
memory performance does not ensure Grounding Awareness. More surprisingly, providing more complete user information can increase execution of unacknowledged proposals, while persistent
memory can cause
agents to treat these proposals as if they were user instructions. We trace these failures through
memory writing,
retrieval, and action selection. Finally, we introduce Grounding Layer, a lightweight, training-free extension that explicitly tracks grounding status and checks consequential actions, substantially improving
agent behavior without modifying models or
memory systems.
Read the full abstract ▾
All authors Jiajun Chen, Xinyue Chen, Yicheng Jiang, Hua Shen
Artificial Intelligence · Cross-listed from cs.LG
ProAS: A Progress-Aware Advantage Shaping Framework for Multi-Turn Agentic Reinforcement Learning
Bowen Ding, Mingda Wan, Yuhan Chen et al. (+7 more)
Reinforcement learning (RL) has emerged as the primary paradigm for optimizing Large
Language Model (LLM)
agents, yet efficient critic-free
algorithms (e.g., GRPO) suffer from a fundamental uniform credit assignment dilemma that broadcasts trajectory-level advantages equally to all actions. Our analysis reveals that this uniformity inherently reinforces structurally invalid actions in preferred trajectories and ignores the downstream cascading influence of early pivotal deviations in unpreferred ones. To resolve this without introducing critic overhead, we propose Progress-Aware Advantage Shaping (ProAS), a plug-and-play framework that leverages a dynamically evolving Agentic Reference Graph (ARG) to achieve fine-grained credit assignment via Masking Unreasonable Incentives (MUI) and Asymmetric Penalty Redistribution (APR). Extensive experiments across Qwen2.5-1.5B-Instruct and Qwen2.5-7B-Instruct demonstrate that ProAS universally elevates mainstream group-based baselines to new state-of-the-art performance, achieving peak success rates of 91.9% and 94.3% on ALFWorld, alongside 66.7% and 70.8% on WebShop, respectively. Furthermore, it accelerates training convergence by up to $2.44\times$ while maintaining strong robustness against imperfect or distilled ARGs. Codes: https://github.com/LINs-lab/ProAS.
Read the full abstract ▾
All authors Bowen Ding, Mingda Wan, Yuhan Chen, Jiyao Yuan, Qi Zhu, Futing Wang, Heyuan Deng, Fei Mi, Lifeng Shang, Tao Lin
Artificial Intelligence · Cross-listed from cs.LG
Model-Based Online Decision Making via Generative Trajectory Planning
Haldun Balim, Yilun Du and Na Li
Online decision-making faces a tension between efficient inference and long-horizon
reasoning. Value-based methods rely on short-horizon bootstrapping that limits the propagation of long-term information, while trajectory optimization methods plan explicitly but must solve a high-dimensional optimization problem at every decision step. We propose Generative Trajectory Planning (GTP), a model-based framework that addresses this tension by learning a generative model as a proposal distribution over action sequences. Instead of optimizing sequences from scratch, GTP samples candidate sequences from a
diffusion prior and refines them using learned dynamics, reward, and value models, enabling efficient trajectory-level
reasoning. Across continuous control and long-horizon manipulation tasks, GTP matches or exceeds state-of-the-art model-free,
diffusion-policy, and model-based baselines, and remains stable in regimes where strong baselines degrade. Our results suggest that generative models are best understood not as policies, but as structured proposal distributions for planning, offering a new perspective on how to integrate generative modeling with model-based decision-making.
Read the full abstract ▾
Artificial Intelligence · Cross-listed from cs.CL
From Relation-Specific Heads to Shared Interventions: Contextual Entrainment in Language Models
Jinghan Zheng and Jiajun Shi
Large
language models rely on
prompt context to perform in-context learning, follow formats, and adapt to local evidence. This ability, however, also raises a robustness concern: models may reuse irrelevant
tokens from earlier context when answering a later query. Prior work identifies this behavior as contextual entrainment and finds
attention heads whose ablation reduces it, arguing that these heads are largely task- or relation-specific. But are the mechanisms behind contextual entrainment truly isolated by relation, or do relation-conditioned searches recover reusable components shared across relations? Using factual-distractor relation-completion
prompts across 19 lexical-semantic subtypes, we identify
attention heads whose ablation reduces entrainment and test how the resulting head sets transfer recovery ability across exact relations, semantic families, and mixed relation distributions. Same-family transfer exceeds different-family transfer, family residuals weaken once globally shared heads are removed, and subtype-specific searches repeatedly rediscover many global heads. Clean-completion and context-use probes further suggest that this shared intervention can reduce harmful copying while preserving ordinary relation completion and simple in-context abilities. Our findings provide a more precise perspective on contextual entrainment and its mitigation: harmful context following is not best addressed by searching for isolated relation-specific
circuits, but by using heterogeneous attack
prompts to reveal reusable intervention targets whose effects are shaped by semantic relation structure.
Read the full abstract ▾
Artificial Intelligence · Cross-listed from cs.CL
SMI: Ontology-Grounded Semantic IDs for Hierarchy-Aware Medical Concept Representation
Ziyang Song, Lia Shen, Yixuan Li et al. (+6 more)
Biomedical
language models (LMs) provide strong representations for clinical tasks, but they often fail to capture the hierarchical structure among medical concepts. This limitation is especially pronounced for rare and fine-grained clinical codes in long-tailed vocabularies, where hierarchy-agnostic LM representations become less informative as concept frequency decreases. We introduce Semantic Medical Identifier (SMI), a training-free framework for constructing ontology-grounded Semantic IDs and hierarchy-aware medical concept representations. For each diagnosis, procedure, or medication concept, SMI defines its Semantic ID as the coarse-to-fine path through an expert-defined medical ontology. SMI then aggregates frozen biomedical LM
embeddings across the corresponding ontology levels to form hierarchy-aware medical concept representations that preserve coarse-to-fine clinical semantics. The base SMI uses uniform aggregation across ontology levels, while a level-weighted variant selects global level weights on validation data. Our theoretical analysis shows that SMI yields more informative medical concept representations than hierarchy-agnostic LM
embeddings, particularly for rare and fine-grained clinical concepts. Experiments show that SMI captures meaningful hierarchical structure and improves clinical relation prediction. On MIMIC-IV, SMI improves mortality prediction, readmission prediction, drug recommendation, and rare drug prediction. On eICU, SMI yields more robust patient representations under cross-hospital distribution shift.
Read the full abstract ▾
All authors Ziyang Song, Lia Shen, Yixuan Li, Qincheng Lu, Weihao Li, Haoran Xu, He Zhu, Xinyu Wang, Yue Li
Artificial Intelligence · Cross-listed from cs.CL
Known-Voice Cloning and the Limits of Personality Transfer for Character Design
Johanna Magdalena Kuch, Maud van Lier, Daksitha Senel Withanage Don et al. (+3 more)
Recent advances in zero-shot voice cloning systems make it possible to reproduce a recognizable voice from a short recording and attach it to arbitrary virtual characters. Prior work suggests that familiar vocal identity can carry over associated personality impressions to new targets such as
robots or virtual characters. However, these works have mainly examined familiarity and personalization effects rather than directly testing whether a cloned familiar voice systematically transfers personality impressions to an embodied virtual character. Such transference raises the concern that cloned voices could be used to manipulate people into engaging with virtual characters as capable social actors. Across two studies, we investigated whether personality impressions transfer when a virtual character speaks with the cloned voice of a familiar person. Study 1 examined this question in a real-world university setting using the cloned voice of a lecturer familiar to the participating students. Study 2 examined the same question online using voices of social media content creators and members of their communities. Drawing on the concepts of identity activation, trait transfer, and person reconstruction, we show that evidence of personality transfer was limited in both studies. Observed effects were selective and did not generalize across traits or study contexts. In the online study in particular, recognizable voice identity cues activated expectations about the speaker but did not reconstruct the broader person impression. Qualitative responses further indicated that judgments were frequently shaped by perceived artificiality and by whether voice and appearance fit together. Our findings moderate common concerns that cloned voices straightforwardly manipulate people into perceiving artificial characters as capable social actors.
Read the full abstract ▾
All authors Johanna Magdalena Kuch, Maud van Lier, Daksitha Senel Withanage Don, Patrick Gebhard, Tanja Schneeberger, Elisabeth André
Artificial Intelligence · Cross-listed from cs.LG
Kwai Summary Attention V2 Technical Report
OneRec Team, Chenglong Chu, Guohong Mu et al. (+30 more)
Summary-
token architectures such as Kwai Summary
Attention (KSA) compress distant context into learnable Summary
token per chunk, and hybrid variants interleave these layers with full
attention to retain exact access to the original text. In such hybrid KSA-Full architecture, the interleaving inserted summary
tokens are served as a streaming flow role across different layers. Although effective, however, it also introduce some bottlenecks: (1) the different layers summary compression chunk must be same, therefore hard to reach a flexible KV Cache compression, e.g., different layer set different summary compression chunk window. (2) in the full-
attention layers, the summary
token introduced additional computation cost, but such layer ignore them and still pay quadratic prefilling, therefore unfriendly to the computation side. Motivated by them, we developed the KSA V2, which not only achieves a more flexible KV cache
memory compression but also reduces the computation cost in Full
attention layer. Specifically, we found that the summary
tokens can serve two roles: the KV
memory and sparse indexer.
Read the full abstract ▾
All authors OneRec Team, Chenglong Chu, Guohong Mu, Guowang Zhang, Hao Peng, Hui Wang, Jian Liang, Jiangxia Cao, Jiao Ou, Jiaxin Deng, Lu Ren, Muhao Wei, Ruiming Tang, Tianle Xu, Wenwu Ou, Xingmei Wang, Zhixin Ling, Zhiyuan Liang, Ziqi Wang, Chaoyi Ma, Haixiao Gao, Kuo Cai, Renpu Liu, Ruochen Yang, Wenlong Xu, Xinqi Jin, Yan Sun, Yifan Wang, Yingzhi He, Yiyu Wang, Yusen Huo, Zhaojie Liu, Zihan Zhou
Artificial Intelligence · Cross-listed from cs.SD
AMemOps: Evaluating Memory Updates in Long Audio Streams
Peiwei Chang, Li Zhang, Mark D. Plumbley et al. (+2 more)
Audio
memory must recognise recurring sources across acoustic changes while keeping their attributes up to date. Overall accuracy does not reveal how these demands interact. We introduce Audio
Memory Operations (AMemOps), a
benchmark for latest-attribute
retrieval, preservation of unchanged attributes and absent-source rejection in long audio streams. Paired controls vary recordings, training identities and earlier events to examine their effects on updating. Training on other recordings of the test instruments increases a reference delta-rule
memory's update advantage over matched
attention by 13.4 percentage points. At
retrieval delays of 120\,s, matched histories separate benefits of consistent repetition from costs of conflicting earlier values, while additive-
memory controls identify benefits of erasure.
Read the full abstract ▾
All authors Peiwei Chang, Li Zhang, Mark D. Plumbley, Philip J.B. Jackson, Wenwu Wang
Artificial Intelligence · Cross-listed from cs.LG
Tiny Recursive Message Passing
Tommaso Salvatori, Matteo Forasassi, Alexis Toumi et al. (+1 more)
Message passing neural networks have shown remarkable performance in algorithmic and constraint
reasoning, due to their ability to encode the structure of the task directly in the architecture. Recent work has shown that better results can be achieved using recursive
transformers trained with deep supervision and truncated recursions, which make extremely deep loops stable to train, and equipped with test-time procedures such as discrete deduction
lattices. In this work, we apply the same procedures to message passing networks, and show that the two ingredients are complementary, and lead to parameter efficient models: we obtain about $99\%$ test accuracy on Sudoku-Extreme with a model of $16$k parameters that trains in about $10$ minutes on an H100, or in $9$ hours on a CPU. We further show that the cost of truncating gradients grows with the number of rounds information needs to cross the graph, making the choice of wiring a trade-off between robustness to truncation and transfer across problem sizes. We conclude by showing that our models transfer zero-shot to larger instances than those seen during training, outperforming popular neural baselines on N-queens and Boolean satisfiability with a fraction of their parameters.
Read the full abstract ▾
All authors Tommaso Salvatori, Matteo Forasassi, Alexis Toumi, Giovanni de Felice
Artificial Intelligence · Cross-listed from cs.CV
ST-GRU PredRNN-V2: A GRU-Based Reformulation of PredRNN-V2 for Video Prediction
Augusta van Haren, Hanna Hoogen, Luca Pattavina et al. (+1 more)
Video prediction requires models to capture complex spatiotemporal dependencies and to retain relevant information over long sequences. PredRNN and its successor PredRNN-V2 address this challenge with spatiotemporal Long Short-Term
Memory (ST-LSTM) units that maintain two separate
memory states, at the cost of a relatively complex recurrent building block. In this work, we investigate whether the simpler gating of Gated Recurrent Units (GRUs) can be transferred to spatiotemporal predictive learning. To this end, we develop a spatiotemporal GRU (ST-GRU) unit as a GRU-based alternative to the ST-LSTM. The ST-GRU replaces the ST-LSTM's three gates per
memory state with two GRU-style gates (update and reset), and merges the temporal
memory state into the hidden state, which reduces the number of parameters by about 5%. We incorporate the ST-GRU into PredRNN-V2 and evaluate the resulting model on the Moving MNIST
video prediction
benchmark, following the original experimental protocol. Averaged over ten predicted frames, the ST-GRU-based PredRNN-V2 achieves an MSE of 55.1, an SSIM of 0.883, an LPIPS of 0.065 and a PSNR of 19.53dB. It does not surpass the original ST-LSTM-based PredRNN-V2 on MSE, SSIM or PSNR, but it achieves a lower LPIPS and performs on par with the earlier ST-LSTM-based PredRNN, while using a simpler recurrent
cell. These results suggest that ST-GRUs are a promising building block for spatiotemporal predictive learning and motivate further work on their design and integration.
Read the full abstract ▾
All authors Augusta van Haren, Hanna Hoogen, Luca Pattavina, Quincy Rommen
Artificial Intelligence
To Align or Not To Align: Check Your COMPASS Before You Train
Fabian Gröger, Petar Damjanović and Maria Brbić
Combining pretrained models is a common strategy for building AI systems, whether by aligning a
vision encoder with a
language model or merging fine-tuned checkpoints. However, finding compatible pairs typically requires costly empirical search, as compatibility is known only after
alignment or merging. We propose COMPASS (COMPatibility ASsessment Score), a theoretically motivated diagnostic that estimates whether a model pair will be compatible prior to
alignment or merging. Building on the observation that representational convergence is most robust at the level of local neighborhoods, COMPASS decomposes model compatibility into two cheap-to-compute axes: a pair-specific ease, defined as the $k$-NN overlap between the two encoders, and a paradigm-level ceiling, capturing the maximum overlap their training paradigms allow. Together, these two axes provide a ranking of candidate model pairs and prescribe how to act on each pair: (i) prioritize pairs that already agree and whose paradigms are compatible, (ii) invest additional compute on pairs with compatible paradigms but low initial agreement, (iii) switch the pretraining paradigm when it cannot support sufficient agreement, or (iv) deprioritize pairs that score poorly on both axes. We apply COMPASS to $72$ cross-modal
alignment experiments and $392$ task-arithmetic model merges and show that compatibility estimates by COMPASS strongly correlate with downstream success, achieving average Pearson correlation of $0.91$ on cross-modal
alignment tasks and $0.70$ on model merging.
Read the full abstract ▾
Artificial Intelligence · Cross-listed from cs.SE
DTScribe: Generating Models for Digital Twin Reporting using Local Large Language Models
Kérian Fiter, Angelica Portocarrero and Bentley Oakes
The development of a digital twin brings together many stakeholders and technical components. We have previously investigated the reporting of the conceptual architecture for these digital twins as an artifact for discussion. However, we find that the activity of modeling the digital twin for visualization and consensus-building can be quite tedious and time-consuming. In this article, we explore the automation of model generation using large
language models, targeting the model required for digital twin reporting. Our approach is designed to ingest digital twin documents and output both the conceptual architecture visualization and a reporting page for that digital twin. We first extract from the input the key details aligning with our twenty-one-characteristic Digital Twin Description Framework described in previous work. Then, we generate a description model in the Ontological Modeling
Language, representing the digital twin as a constellation of models, data, enablers, and services. We utilize refinement loops with large
language models acting as judges and ontological
reasoning to increase artifact quality as measured by our metrics. We investigate how architectural and parameter choices affect artifact quality, such as how refinement loops enable local large
language models to produce sufficiently high-quality results comparable to larger models. We also detail insights and lessons learned useful for practitioners, such that they can integrate our techniques to accelerate their digital twin reporting or for other model generation tasks. These insights include the use of medium-sized local models (around 20 billion parameters), the avoidance of large
language models as judges and clustering for characteristic extraction, and the recommendation of at least two reasoner loops for Ontological Modeling
Language refinement.
Read the full abstract ▾
Artificial Intelligence · Cross-listed from cs.LG
Geometric Calibration of Supervised Fine-Tuning
Jianzhu Bao, Jia-Chen Gu, Qingyuan Liu et al. (+4 more)
Recent studies investigate the weight geometry through its spectral structure, which characterizes how the weights transform different input directions. Within this view, supervised fine-tuning (SFT) is shown to induce dense updates that perturb the spectral geometry of pre-SFT weights along their principal directions, potentially disrupting the model's prior knowledge. These observations motivate using the geometry of pre-SFT weights as a reference for refining learned SFT updates. We study this perspective as
Geometric Calibration and develop SPACE Subspace Projection for Adapted Checkpoint Enhancement), a lightweight post-hoc method that selectively reverts SFT update components within the principal subspace of the pre-SFT weights while preserving those in its orthogonal complement. SPACE is a data-free, training-free method that requires only the pre- and post-SFT checkpoints. We evaluate SPACE on Qwen2.5, Qwen3, Qwen3.5, Llama-3.1, and Olmo-3 models ranging from 2B to 14B, across mathematical, coding, and medical SFT, as well as several SFT variants. Overall, SPACE mitigates out-of-domain (OOD) capability regression while further improving in-domain (ID) performance over the uncalibrated SFT checkpoints. For math SFT, SPACE yields average ID/OOD gains of 2.85/2.26 points on Qwen2.5-7B, and 2.41/1.52 points on Qwen3-14B-Base, respectively. These results highlight how the
geometric relationship between SFT updates and pre-SFT weights shapes the adapted model's capabilities and, more broadly, motivate geometry-aware parameter updates for continual model improvement.
Read the full abstract ▾
All authors Jianzhu Bao, Jia-Chen Gu, Qingyuan Liu, Haozhen Zhang, Kai-Wei Chang, Nanyun Peng, Wenya Wang
Artificial Intelligence · Cross-listed from econ.TH
Falsification-proof non-market allocation mechanisms
Eduardo Perez-Richet and Vasiliki Skreta
Relying on manipulable characteristics to target the allocation of resources is fraught with difficulties. Naively optimal rules lead to falsification and allocative inefficiency. Optimizing the rule for allocative efficiency while accounting for falsification typically induces falsification, at a heavy cost to the
agents. Moreover, falsification generates negative externalities. We characterize optimal falsification-proof rules under a broad range of falsification technologies. We also provide an honesty condition under which they solve the unconstrained problem of a planner who maximizes a weighted sum of allocative surplus and
agents' welfare.
Read the full abstract ▾
Artificial Intelligence
Two Routes to Interoceptive Prioritisation: Dissociating Likelihood Precision and Preference Reweighting in Active Inference
St John Grimbly, Nicolas Kuske, Evert A. Boonstra et al. (+6 more)
An embodied
agent must prioritise among competing bodily needs. In active inference, it can respond to an urgent need by increasing the precision of information about that need or by increasing its preference for outcomes that relieve it. We compare these two interventions, separately and together, in a foraging gridworld with three lethal needs, and ask how each changes survival, perception and the body model the
agent learns. Precision sharpens interoceptive posteriors and improves survival. Preference reweighting keeps bodily states further from their lethal boundaries and changes the states from which the
agent learns, so its learned model covers fewer need levels, with lower quality across all levels but slightly higher quality on the levels it visits. An experiment on 72 further layouts does not establish the predicted preference benefit on difficult layouts, but both interventions improve average survival there and their interaction is positive. Survival alone therefore gives an incomplete account of how a model prioritises needs and learns about its body.
Read the full abstract ▾
All authors St John Grimbly, Nicolas Kuske, Evert A. Boonstra, Charel van Hoof, Rowan Hodson, Benjamin Rosman, Ryan Smith, Mark Solms, Jonathan P. Shock
Artificial Intelligence · Cross-listed from cs.CL
Knowledge Graphs as Files: An Agent Harness Where Tool Creation and Tool Use Co-Evolve
Yuyang Bai, Yu Wang and Yu Zhang
Searching text-rich knowledge graphs requires following relations while examining textual evidence. Predefined graph interfaces constrain what evidence
agents can access, even when they write code to combine tool calls or process retrieved subgraphs. We propose direct knowledge graph interaction (DKGI), a harness in which
language model
agents create graph access tools directly over complete node and edge files. The
agent writes Python code to follow relations, combine node sets, and select textual evidence, interleaving execution with shell search and file inspection. This lets
agents process large node sets in bulk while returning selected text for the model to assess. Execution feedback guides tool creation and revision, while the evolving tools determine what evidence the
agent gathers next. On
benchmark datasets spanning three domains, training-free DKGI exceeds the strongest published baseline by at least 6.7 Hit@1 points on every graph, even with a minimal harness. Controlled ablations show the benefits of multi-turn programming over shell-only access and predefined graph tools, while trajectories show
agents revising code to extend searches or address errors. These findings highlight the value of allowing tool creation and tool use to co-evolve during graph search. Our code and data are available at https://github.com/leopoldwhite/DKGI.
Read the full abstract ▾
Artificial Intelligence
Can an Agent Learn to Predict Its Own Failure? Outcome-Grounded Confidence Calibration in LLM Agents
Ashutosh Hathidara
Controlling an acting LLM
agent means deciding, at each step of a long-horizon task, whether to trust its next action, abstain, or hand off to a human. That decision needs a per-step confidence signal predicting whether the current rollout will eventually succeed. The
agent's own self-reported verbalized confidence is the natural candidate, but existing reinforcement learning methods calibrate it against the correctness of a decision at the current step. We propose a method to calibrate it against the eventual outcome of a multi-step episode: under GRPO, the group of rollouts for a task supplies success labels, and a proper scoring rule scores each step's confidence against the realized outcome. In offline RL on recorded trajectories of a Qwen2.5-7B
agent in ALFWorld, TextWorld, and WebShop environments, the AUROC of step confidence against episode outcome rises from near chance to 0.68-0.85, and the top-10% of steps the
agent is most confident about succeed ~4.3x more often than average. In online RL on the live ALFWorld
agent, four calibration rewards improve early-step AUROC, expected calibration error goes from 0.42 to below 0.12, and selective execution improves over both the untrained
agent and a format-only RL control, without degrading task success. We also introduce an evaluation protocol that separates episode depth, task difficulty, and attempt-level foresight, which shows where the gain lies: the trained
agent recognizes from its first step which tasks it will fail, while telling a succeeding attempt from a failing one at the same task is harder, with some early-window signal emerging on TextWorld.
Read the full abstract ▾
Artificial Intelligence · Cross-listed from cs.LG
Jason Ken Adhinarta, Raina Wu, Ashutosh Sharma et al. (+5 more)
Self-
distillation is an emerging approach that trains a model to imitate its own predictions when conditioned on privileged information, such as correct answers or environmental feedback. Recent methods have reported better sample efficiency and generalization than reinforcement learning (RL), but it remains unclear whether these gains extend beyond in-distribution evaluations and whether they stem from privileged
distillation. In this paper, we revisit two representative methods, Self-
Distillation Policy Optimization (SDPO) and On-Policy Self-
Distillation (OPSD), across code, science, tool use, and math
reasoning tasks. We find that their reported gains do not consistently generalize to unseen problems and are largely explained by overfitting or transferring existing
reasoning behavior, rather than effectively using privileged information. Moreover, self-
distillation gradients show weak
alignment with policy gradients that directly optimize expected reward, questioning whether they provide an effective direction for policy improvement. Our experiments across Qwen3 models (1.7B-14B) and OLMo-3-7B suggest that effectively leveraging privileged information to improve
reasoning remains an open challenge. Our code is available at https://github.com/sleepypandu/does-sd-improve-
reasoning
Read the full abstract ▾
All authors Jason Ken Adhinarta, Raina Wu, Ashutosh Sharma, Avinash Balakrishnan, Rameswar Panda, Polina Golland, Yoon Kim, Zhang-Wei Hong
Artificial Intelligence · Cross-listed from cs.CR
Nudge the Judge: Uncovering Program Decision Hijacking in Real-World Jev Applications
Yang Luo, Tiantian Ji, Zifeng Kang et al. (+4 more)
System One models such as Jev change how software makes decisions by mapping open-ended semantic contexts into structured judgments that directly drive program execution. However, this paradigm introduces a new security risk, which we term Jev Program Decision Hijacking (JPDH): attacker-controlled contexts can redirect program decisions through schema-valid Jev judgments. Existing evaluations focus primarily on model-level attacks or predefined injection channels, leaving attack reachability and downstream security consequences in real-world Jev applications underexplored. In this paper, we present JevProbe, the first automated framework to evaluate JPDH by tracing
adversarial influence from reachable context surfaces through program-consumed decisions to downstream security consequences in real-world Jev applications. Our key insight is that application logic maps diverse Jev judgments to observable decisions, enabling systematic, program-grounded attack evaluation. Across 228 popular real-world Jev applications and 2,273 test cases, 81.5% exhibit decision flips, of which 98.6% lead to potential downstream security impacts. Compared with the strongest adapted baseline, JevProbe improves macro-average Decision Flip Rate (DFR) by 50.4 percentage points, while existing defenses offer limited protection. We further construct JevProbeBench to support systematic evaluation of JPDH in real-world Jev applications. We have responsibly disclosed our findings to Jev application maintainers, pending their responses. The source code of JevProbe and the JevProbeBench
dataset are publicly available on GitHub (https://github.com/xiaoxiaohg8d/JevProbe-).
Read the full abstract ▾
All authors Yang Luo, Tiantian Ji, Zifeng Kang, Yucheng Wang, Yong Liu, Zhiyi Yin, Xinran Liu
Artificial Intelligence · Cross-listed from cs.CL
Jev-ToM: Evaluating Theory of Mind in Typed Decision Models
Kawai Chung, Chunkit Chan, Yuxuan Liu et al. (+9 more)
Dialogue and collaborative assistants repeatedly judge what a partner knows, believes, or intends. Typed decision models offer a potential alternative to generative large
language models for bounded Theory of Mind (ToM) selection. We present Jev-ToM, comparing five typed decision models and seven generative baselines on 11,117 English text questions from nine ToM
benchmarks across six capability groups. TypeSafe Jev reaches 73.2% source-balanced accuracy including native controls and ranks first or second across all six capabilities, leading beliefs, goals/desires/intentions, and emotions. Other typed models show distinct strengths across tasks. Separate resource comparisons suggest that Jev could reduce the cost and waiting time of repeated ToM selection. Confidence-based selection remains
benchmark-dependent, and supplied reasons can repair errors or overturn correct answers depending on their attachment. Candidate expansion reveals distinct selection and output-validity problems. Related-question cues account for most of Jev's downstream gains, with task-dependent effects from predicted states. The findings motivate testing Jev as a ToM selection component in assistants, with task-specific confidence calibration, candidate design, and output validation as integration priorities.
Read the full abstract ▾
All authors Kawai Chung, Chunkit Chan, Yuxuan Liu, Yauwai Yim, Xi Yang, Wenbin Hu, Huihao Jing, Chiu Yeung Tang, Changxuan Fan, Ki Sen Hung, Tsun On Kwok, Haoran Li
Artificial Intelligence · Cross-listed from cs.CR
DPOison-VL: Clean Preference Poisoning in Vision-Language Direct Preference Optimization
Chengtao Jian and Kai Yang
Direct Preference Optimization (DPO) has become a widely adopted approach for aligning
vision-
language models (VLMs), yet the security of
multimodal preference data remains largely unexplored. Existing preference poisoning attacks either corrupt explicit supervision signals or require visible triggers, making them detectable under human inspection. We introduce DPOison-VL, the first clean preference poisoning attack against VLM-DPO, which manipulates model
alignment solely through imperceptible perturbations on a small fraction of training images, while leaving all
prompts, responses, and preference annotations semantically intact. We formulate poison construction as a bilevel optimization and derive a gradient
alignment surrogate that efficiently steers image perturbations toward the attacker-specified preference. We further introduce a preference displacement score for principled poison selection with provable first-order optimality under a fixed poison budget, and derive a sufficient condition for preference inversion. Extensive experiments across diverse attack tasks, different VLM architectures and scales, and baseline comparisons demonstrate that DPOison-VL achieves up to 98% ASR with only 2% poisoned data and exhibits cross-model transferability, while largely preserving clean accuracy and general utility. Our findings reveal a critical vulnerability in VLM-DPO, where imperceptible image perturbations in preference training data can silently redirect
alignment.
Read the full abstract ▾
Artificial Intelligence
Beyond Answer Voting: Fine-Grained Correction of LLM Reasoning through Graph Alignment across Sampled Traces
Haotian Liu, Yingjian Chen, Xiaohang Xu et al. (+4 more)
Large
language models increasingly tackle tasks that require multi-step
reasoning, where errors in intermediate claims can affect the final answer. Self-consistency (SC) improves
reasoning by voting over sampled final answers, but its vote leaves intermediate disagreements unresolved. To address this problem, we introduce DDRG (Disagreement-Driven Verification Guided by
Reasoning Graphs), a framework for selective correction from a frozen bank of
reasoning traces. Our insight is that graph
alignment can expose claim-level disagreements and provide correction evidence without expanding the sampled trace bank. DDRG builds source-grounded claim graphs and sends localized conflicts to an answer-blind verifier. A verifier-supported refutation that reaches the majority answer's supporting path triggers one focused repair; otherwise, DDRG returns the exact SC vote. Experimental results across five
benchmarks show that native zero-shot DDRG improves macro-average accuracy over direct answering by 8.65 and 15.83 percentage points for Qwen2.5-7B and Llama3.1-8B. It also exceeds CoT, the evaluated GoT and GoV question-answering adaptations, and SC on this macro metric. DDRG links local verification to selective correction without enlarging the initial candidate bank.
Read the full abstract ▾
All authors Haotian Liu, Yingjian Chen, Xiaohang Xu, Lang Huang, Yusuke Iwasawa, Yutaka Matsuo, Irene Li
Artificial Intelligence · Cross-listed from cs.CV
LLM2R: From Large Language Models to 3D Reassemblers
Zhouhan Zhong, Qiuhong Shen, Xingyi Yang et al. (+2 more)
3D part assembly aims to reconstruct a complete shape from a set of unassembled parts by predicting a rigid SE(3) transformation for each part. Because the placement of a part is rarely determined by its own geometry, the task demands
reasoning over the entire part set, and existing methods, which rely on pairwise relations or iterative generation and verification, scale poorly to objects with many parts. We present LLM2R, which casts part assembly as sequence modeling with a pretrained decoder-only large
language model (LLM): a Point-BERT encoder and a Q-Former compress each part into a block of
geometric tokens, the blocks of all parts are concatenated, and a causal LLM adapted with Low-Rank Adaptation (LoRA) processes the sequence in one forward pass. Causal
attention, however, lets the hidden states of a part observe only the parts that precede it. LLM2R therefore appends a block of learnable summary
tokens after the last part: these
tokens follow every part, so their states aggregate the whole assembly at a constant
token cost, and a part-aware pose head fuses each part's states with them before predicting its rotation and translation. One architecture thus handles semantic and
geometric assembly and predicts all poses in a single non-autoregressive pass. LLM2R achieves the highest part accuracy and success rate on all three PartNet categories, the best rotation error, translation error, and part accuracy on PartNeXt, and the lowest translation error and fastest inference on Breaking Bad, where only a fracture-specialized method is more accurate. Its advantage grows with the number of parts: at 100 parts it exceeds GARF by 33.5 points in part accuracy on Breaking Bad and Assembler by 29.1 points on PartNet, and it transfers zero-shot to unseen categories with more than 16 points higher part accuracy than the strongest of three baselines.
Read the full abstract ▾
All authors Zhouhan Zhong, Qiuhong Shen, Xingyi Yang, Jinyi Xu, Xinchao Wang
Artificial Intelligence
AgentStripes: An Evidence-Graded Maturity Model for AI Agents
Nutan Kumar Naik, Vijay Prasad Poudel, Aditya Kumar Saroj et al. (+2 more)
Benchmarks report what an Artificial Intelligence (AI)
agent can do on a task suite, but not whether one released version has been engineered with the controls that make it reliable, safe and observable, or how strongly that is shown. This paper proposes AgentStripes, an evidence-graded maturity model for AI
agents. It defines levels (0 to 7 stripes), process areas (12 areas holding 40 capabilities), three gates (basic, safety, proof) that hold an
agent at a level however high its score, and evidence grades E0 to E5. Each applicable capability receives a rating from 0 to 3 and a separate evidence level; these are combined into a credit, aggregated into a severity-weighted score and converted into stripes that the gates limit. Instruments gather evidence,
language-model judges interpret it under explicit criteria, and the arithmetic is deterministic. The model is compared with standards at topic level without a claim of conformance, and an unexecuted validation protocol is specified. An empirical evaluation (exploratory, non-confirmatory) of 30 public Python
agent repositories, with static evidence only, is a demonstration and not a validation: static evidence caps every final stripe at 2, 26 repositories end at one stripe and 4 at two, and this distinction is unreliable, since re-judging changed the final stripe of 4 of 22 repositories (engine, dossier and panel changed), whereas none of 8 panel-only repeats, all at stripe 1, changed. The parameters are expert conventions without outcome based calibration, and the stripes have not been shown to predict reliability or safety.
Read the full abstract ▾
All authors Nutan Kumar Naik, Vijay Prasad Poudel, Aditya Kumar Saroj, Saurav Samantray, Abhishek Patel
Artificial Intelligence
Rongtian Ye
Point-in-time (PIT)
agents can recall post-cutoff facts even when their external evidence respects a historical cutoff. We examine whether this access predicts evidence gathering in 48 historical company-comparison tasks. Independent closed-book revenue probes identify the better-remembered company for each model on tasks with unequal recall, fixing the predicted direction before
agent runs. We then compare paired runs with real or anonymized company identities while keeping external evidence PIT-valid. Both Qwen and DeepSeek recalled future facts, but neither supported the prespecified shift toward the better-remembered company. Qwen's four allocation estimates on 28 unequal-recall tasks were negative, with descriptive intervals spanning zero. A prospective DeepSeek replication showed an inverse pattern across all four metrics on 20 such tasks, with intervals below zero. These estimates remained negative after each company, pair, and future-fact omission, and all four post-hoc max-T-adjusted sign-flip p-values were below .05. In a prospectively specified follow-up, explicit future facts produced small estimated shifts away from the supplied-fact company, with intervals spanning zero, while final choices shifted toward it. The mechanism behind the natural inverse pattern remains unresolved. These findings support separate assessments of
memory accessibility, search allocation, and final outputs alongside cutoff checks on external evidence.
Read the full abstract ▾