Relation
-
GaugeVLM: Structuring Spatial Supervision with Measured Geometric Interventions
2026-09-29 16:26:13Hongbo Wang, Zihan Lin, Wenkui Yang, Shiran Ge, Yuang Ai, Jie Cao, Huaibo Huang, Ran HeAbstract
Vision-language models (VLMs) can contradict themselves across views of the same spatial relation and fail to respond when that relation changes. Addressing these failures requires supervision that captures error magnitude and geometric dependencies across observations, both of which remain implicit in training on individual answers or ordinal preferences. Therefore, we introduce GaugeVLM, which makes this structure explicit through controlled object and camera interventions in explicit 3D scenes, producing linked observations with measured differences between spatial relations and shared truths across views. To translate this structure into learning signals, its core objective, GaugeDPO, converts measured errors into preference margins, directly supervises correct canonical rankings across views, and links intervention-induced answer-odds contrasts to measured relation changes with view-specific scales. Our analysis bounds canonical prediction error and establishes that the cross-view and intervention constraints can be jointly satisfied. Empirically, GaugeVLM improves all 10 established spatial metrics over supervised fine-tuning across three VLM backbones, with the main 7B model gaining 15.0 and 18.9 percentage points on MSMU distance and QSpatial+, respectively. These gains also extend to autonomous driving and embodied reasoning, demonstrating the robust generalization across domains.
Abstract (translated)
URL
https://arxiv.org/abs/2609.38285
PDF
-
Does Local Video Understanding Transfer Across Encounters? The EgoGears Benchmark
2026-09-29 16:20:01Yuedong Tan, Lei Qi, Yu Liu, Di Wen, Ruiping Liu, Xiaoye Wang, Yufan Chen, Junwei Zheng, Chengzhi Wu, Chen Zhang, Zhihang Chen, Haiwen Sun, Zongwei Wu, Radu Timofte, Danda Pani Paudel, Kunyu PengAbstract
Embodied systems must make knowledge acquired during one encounter usable in another despite changes in viewpoint, motion, and illumination. Yet aggregate cross-video accuracy conflates failures of local perception with failures to preserve observation identity, establish correspondence, and compose evidence, obscuring whether local video understanding actually transfers. We introduce EgoGears, a complementary single- and multi-video benchmark designed to diagnose this transition. It contains 567 single-video and 1,487 multi-video questions derived from 126 human-collected egocentric recordings covering 39 outdoor routes. Repeated traversals across movement speeds and lighting conditions ground comparisons in shared physical environments; 531 questions require alignment across independent recordings. Single-video questions measure the local visual, spatial, and motion evidence available to a model, while multi-video questions test whether evidence remains bound to the correct observation and can be composed into consistent route relationships. We report 29 single-video and 31 multi-video MLLM configurations across six model families in the main leaderboard. Among the 20 configurations evaluated comparably on both splits, every model performs worse on multi-video questions, with a mean decrease of 22.5 percentage points, and the gap persists when answer format and scoring are held fixed. The gap is not explained simply by additional videos or recording boundaries. The central bottlenecks are observation--evidence binding and ordered route-state tracking. The code and benchmark are publicly available at this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2609.37938
PDF
-
Generated Query Expansion Still Helps Strong Sparse Retrieval: A Controlled Study with SPLADE-v3
2026-09-29 16:07:06Ryan C. Barron, Cade W. Trotter, Maksim E. Eren, Kim {\O}. Rasmussen, Liz D. Miller, Benjamin J. MiglioriAbstract
Scientific queries are often brief, while relevant papers use specialized vocabulary. Generated query expansion can bridge this mismatch, but earlier work suggests that its value shrinks as the underlying retriever becomes stronger. We test the four generated formats of term lists, a pseudo-document, multiple pseudo-references, and corpus-steered text all together with SPLADE-v3 on NFCorpus, TREC-COVID, and SciDocs. Every condition searches the same frozen document index and follows the same query-side integration rule and 256-dimension budget, isolating the effect of the added content. All twelve method-collection comparisons improve aggregate nDCG@10, with best relative gains of 4.81%, 8.92%, and 9.47%. Eleven remain significant after Holm correction. The gain persists in 103 of 114 interpolation settings, including every setting that assigns at least 30% of the mixture weight to the original query. Shuffled-text and non-contextual lexical-bag controls also remain above baseline in all 24 aggregate comparisons, showing that the added vocabulary carries most of the benefit. A corpus-induced typed concept graph, by contrast, produces no consistent gain, and its relation, depth, validation, random, and gating controls do not rescue it. Generated vocabulary can therefore complement a strong learned sparse retriever, provided that the original query remains strongly represented.
Abstract (translated)
URL
https://arxiv.org/abs/2609.37911
PDF
-
ReCAP: Retrieval-Guided Capability Reuse for Multimodal Continual Instruction Tuning
2026-09-29 15:56:11Tao Hu, Zhinuo Zhou, Xialiang Tong, De-Chuan Zhan, Da-Wei ZhouAbstract
Multimodal continual instruction tuning (MCIT) aims to enable multimodal large language models to acquire new capabilities from sequential tasks while preserving previously learned knowledge. Existing methods primarily mitigate catastrophic forgetting by constraining parameter updates or separating task-specific adaptations. However, continual adaptation can also benefit from external knowledge that provides domain-specific information and reusable reasoning patterns for solving diverse instructions. For example, to answer "How many red cubes are to the left of the sphere?", domain knowledge can provide relevant concepts about objects and spatial relations, while reasoning knowledge can specify ordered operations such as object recognition, spatial filtering, and counting. Despite this potential, how to leverage external knowledge for continual adaptation remains largely unexplored in existing MCIT methods. To this end, we propose ReCAP, a retrieval-guided framework that leverages external knowledge to guide capability reuse during continual adaptation. At each continual stage, ReCAP uses external search and an LLM to incrementally build a knowledge base of domain, reasoning, and format knowledge based on the current-stage training data. For each instruction, retrieved domain knowledge guides generation, while retrieved reasoning knowledge selects and orders capability modules to form an instance-specific capability path. As these capability modules are reused across stages, subsequent adaptation can overwrite previously learned parameters. To enable stable cross-stage reuse, ReCAP introduces adaptive subspace recycling, which parameterizes reusable capability modules with shared bases and stage-specific cores, protects historically important directions while recycling residual capacity. Extensive experiments on MCIT benchmarks show that ReCAP achieves SOTA performance.
Abstract (translated)
URL
https://arxiv.org/abs/2609.37889
PDF
-
LLMs are not stochastic parrots: Evidence for meaning-mediated abstraction from conlang-like tasks
2026-09-29 15:46:02Julia Witte Zimmerman, Calla G. Beauregard, Tabia Tanzin Prama, Parisa Suchdev, Kathryn Cramer, Elisabeth KollrackAbstract
The strong version of the stochastic parrot argument claims that, although large language models (LLMs) may exceed rote regurgitation, they cannot move beyond statistical pattern matching into abstraction or reasoning, remaining ontologically near the lower bound of pattern reuse despite producing alluringly fluent text. We test this hypothesis using conlang-like tasks. Several LLMs are given only natural-language descriptions of fictional languages that subvert prominent superficial patterns in training data by combining statistically uncommon and unattested features. Crucially, no example outputs are given. We argue that if the models exhibit rule-following behaviour, they cannot be relying solely on superficial statistical patterns; such patterns often work against the correct output. Instead, successful performance requires representations of the constraints specified in the prompt. Across three complementary task families, models systematically move in the meaning-predicted direction: they distinguish prompt exposure from instructed use, alter semantic relationships in response to novel constraints, and sometimes produce exact matches to complex translation answer keys. Although performance varies across the spectrum of models used, these results provide evidence for meaning-mediated abstraction in LLMs and refute the strong stochastic parrot hypothesis. Our work shows that, under appropriate architectural and contextual constraints, statistical learning can produce meaning-mediated abstractions, although generation remains strongly constrained by superficial plausibility. We discuss implications for model development and for understanding how increasingly abstract representations may emerge from plausible-text-generation objectives.
Abstract (translated)
URL
https://arxiv.org/abs/2609.34187
PDF
-
AnthroDial: Benchmarking LLM Anthropomorphism in Autonomous Social Interaction
2026-09-29 15:43:07Wentao Liu, Xi Chen, Siyu Song, Biao Yuan, Yu Zhang, Zhou Zhuotong, Jingying Zhou, Guohao Feng, Shasha Hu, Tianfu Wang, Shangshang Yang, Haoyang Liu, Youjia Li, Xiaokun Wang, Min Ji, Ji WangAbstract
Large language models (LLMs) are increasingly deployed as social agents, yet credible human-like interaction requires more than fluent responses or persona consistency. Agents must autonomously decide whether, when, and how to communicate while adapting to evolving contexts, goals, and relationships. Existing research, however, lacks a unified approach to enabling, evaluating, and improving such capabilities in continuous, open-ended interaction. We introduce AnthroDial, a unified framework for developing anthropomorphic social agents from three complementary aspects: MindFlow, a lightweight interaction harness that enables autonomous, asynchronous, and adaptive communication through a dynamic Mind Buffer; CAPS-Eval, a theory-grounded framework for evaluating cognitive, affective, and behavioral dimensions of anthropomorphic interaction; and a scalable training paradigm that combines SEEDS for environment expansion with DiAPO for adaptive capability optimization. We further construct evaluation datasets covering everyday communication, game interaction, and long-horizon character interaction. Extensive experiments across diverse models and scenarios demonstrate improved interaction autonomy and naturalness, validate the reliability, discriminativeness, and agreement with human rankings of CAPS-Eval, and confirm the effectiveness of our training paradigm. Together, these components provide a unified framework for developing credible human-like social agents in open-ended interaction.
Abstract (translated)
URL
https://arxiv.org/abs/2609.37853
PDF
-
Which Models Work Well Together? Measuring Heterogeneity for LLM Team Selection
2026-09-29 15:30:53Liangyu Teng, Hengsong Liu, Juncen Guo, Jingyu Zhang, Yang Liu, Jing Liu, Liang SongAbstract
The performance ceiling of an LLM team is constrained not only by individual model capabilities, but also by inter-member error resonance and predictive differences. Although heterogeneous teaming is often observed to be effective in practice, existing approaches lack complementarity metrics that are computable, interpretable, and optimizable, leaving team composition to rely on heuristics. We propose a heterogeneity-driven team selection framework that performs offline profiling to characterize individual capability along with two complementary signals: one captures decorrelation in error patterns to reduce co-failures, while the other measures divergence in predictive behavior to capture strategy diversity. We formulate team selection as a standardized quality--complementarity combinatorial objective and apply an efficient greedy search to select a small team from a candidate pool. Experiments across multiple benchmarks demonstrate that our framework consistently outperforms quality-only baselines under controlled candidate pools and team sizes, establishing reusable selection principles for multi-LLM systems.
Abstract (translated)
URL
https://arxiv.org/abs/2609.38274
PDF
-
MVG-WAM: Multiple View Geometry-Aware World-Action Modeling for Robotic Manipulation
2026-09-29 15:20:59Wenbo Chen, Tianfu Li, Haoxuan Xu, Zhihao Cao, Zhenghan Chen, Zhengming Zhu, Zizhou Luo, Guosheng Yang, Yuan Liu, Lujia Wang, Wen Chen, Haoang LiAbstract
World-Action Models (WAMs) couple visual dynamics with action prediction, bringing the rich priors of pretrained video models to robotic manipulation. However, their multi-view interfaces typically tile images or concatenate tokens, leaving the geometric relationships among synchronized cameras implicit. This makes it harder to connect global scene context with the local geometry required for interaction. We introduce the Multi-View Geometry-Aware World-Action Model (MVG-WAM), which organizes these observations as related projections of one physical world rather than separate images on a canvas. Our model combines an epipolar-constrained global state with view-indexed geometric states jointly inferred from synchronized observations. Camera-aware routing supplies each video region with its corresponding geometric context and the shared global state, explicitly structuring the representation used for action prediction. We further ground the geometry-aware representation in metric scale through multi-horizon future-depth supervision, without requiring depth decoding during action rollout. MVG-WAM achieves average success rates of 99.1% on LIBERO and 92.07% on RoboTwin 2.0, demonstrating competitive performance across both benchmarks. Real-world experiments on Cobot Magic further demonstrate a 91.3% success rate across 150 trials spanning three manipulation tasks.
Abstract (translated)
URL
https://arxiv.org/abs/2609.37793
PDF
-
Geometry-Preserving Human-to-Robot Upper-Body Motion Retargeting from Monocular Video
2026-09-29 15:14:20Xiaoyu Yang, Sen Han, Da Li, Nan WuAbstract
Monocular RGB video provides an accessible source of human demonstrations for upper-body robot motion, yet video-driven human-to-robot transfer remains challenging because body and hand motion are recovered at different spatial scales, human and robot kinematics differ substantially, and fine distal motion is difficult to preserve across embodiments. We present a geometry-preserving motion-retargeting framework that integrates unified body--hand reconstruction with morphology-independent geometric transfer. Frame-wise body estimates, video-level observations, and detailed hand evidence jointly constrain a single differentiable Momentum Human Rig (MHR) state, while transient hand artifacts are repaired in parameter space. The reconstructed motion is represented by arm-segment directions, elbow configuration, relative palm orientation, and bilateral wrist relations, and is realized on the target robot through multi-stage inverse kinematics and robot-specific hand adaptation. Within the broader system, Across-VAM provides video generation, whereas Across-WAM performs human-to-robot motion mapping. The method is evaluated on 16 monocular videos comprising 1,769 source frames, including 10 signing and six reach-to-grasp sequences. Unified reconstruction reduces mean hand reprojection error from 22.36 to 7.21 pixels relative to SAM 3D Body. All 16 retargeted trajectories completed kinematic simulation playback, and representative signing and reach-to-grasp motions were further demonstrated on a physical robot. The results demonstrate a unified pipeline from monocular human video to coordinated upper-body motion on a dual-arm dexterous robot.
Abstract (translated)
URL
https://arxiv.org/abs/2609.37776
PDF
-
Evaluating Multi-Task Morphological Concept Learning for Pulmonary Nodule Malignancy Assessment in 3D CT
Abstract
Morphological characteristics such as spiculation and lobulation play an important role in assessing pulmonary nodules on computed tomography (CT), particularly in relation to malignancy risk. This study examines whether learning radiologist-annotated morphological features together with malignancy risk from lesion-centred 3D CT volumes improves classification performance. The Lung Image Database Consortium and Image Database Resource Initiative (LIDC-IDRI) dataset was used, comprising 3,918 reader-level nodule annotations from 742 patients after excluding indeterminate malignancy ratings. Patient-level splitting was used for training, validation, and testing, with 112 patients and 628 reader annotations in the held-out test set. A single-task 3D convolutional neural network was compared with a multi-task model predicting malignancy risk, spiculation, and lobulation. The single-task model achieved a balanced accuracy of 0.548 and receiver operating characteristic area under the curve (ROC-AUC) of 0.552, while the multi-task model achieved 0.539 and 0.558, respectively. Patient-level bootstrap analysis showed an ROC-AUC difference of 0.005 (95% confidence interval (CI): -0.087 to 0.090) and a balanced-accuracy difference of -0.009 (95% CI: -0.067 to 0.043). The auxiliary tasks were strongly imbalanced and showed limited predictive performance. Overall, including morphological features did not clearly improve malignancy-risk classification, showing the importance of class balance, label formulation, and reader-level annotation structure in multi-task pulmonary CT analysis.
Abstract (translated)
URL
https://arxiv.org/abs/2609.38271
PDF
-
Reader Proficiency Shapes Layer-wise Surprisal Profiles
2026-09-29 14:32:50Akio Hayakawa, Horacio SaggionAbstract
Reading behaviour varies not only with linguistic input, but also with reader proficiency. In this study, we investigate whether the layer-wise relationship between surprisal from large language models (LLMs) and human gaze behaviour differs across readers with different levels of proficiency and across gaze measures. Using eye-tracking data from the MECO L2 corpus, we compare readers with high and low vocabulary proficiency on first-pass gaze duration (FPGD) and total gaze duration (TGD). We quantify the distribution of the predictive power of surprisal across model layers using Predictive Depth. Across 12 tested LLMs, we find that readers with lower vocabulary proficiency tend to show deeper Predictive Depth for FPGD, while this difference is smaller for TGD. Also, TGD itself shows deeper Predictive Depth than FPGD in both proficiency groups. These patterns suggest that where predictive power is concentrated across LLM layers may be related to the timing and breadth of the reading processes captured by different gaze measures, and that this relationship can vary with reader proficiency. Our leave-one-out analysis further shows that the advantage of informative internal layers extends to unseen texts, although the practical improvements in prediction are limited. Overall, our results show that layer-wise LLM surprisal provides a useful perspective on variation in reading behaviour across both reader groups and gaze measures.
Abstract (translated)
URL
https://arxiv.org/abs/2609.37688
PDF
-
Corpus-Guided Dual-Path Propagation for Graph Retrieval-Augmented Generation
Abstract
Graph-based retrieval-augmented generation supports multi-hop retrieval by organizing corpus information into graphs. However, existing relation-free graph retrieval methods rely primarily on query-sentence similarity to search for evidence. This can exclude useful bridging evidence with low query similarity and activate incidental entities unrelated to the reasoning chain. In this paper, we propose a simple and effective approach called NexusRAG, which augments the relation-free Tri-Graph with a corpus-level entity neighborhood structure derived from joint entity co-occurrence and semantic similarity. NexusRAG employs this structure to guide two complementary propagation paths: neighborhood-constrained semantic propagation through sentences identifies the query-relevant entity frontier, while direct structural propagation between neighboring entities expands that frontier to structurally related entities. The propagated entity weights also inform neighborhood-aware passage initialization for Personalized PageRank. Experiments on three multi-hop QA benchmarks and a domain-specific subset of GraphRAG-Bench show that NexusRAG consistently outperforms existing approaches. On the GraphRAG-Bench subset, NexusRAG achieves the highest evidence recall in all question categories, exceeding baselines by 4.2-8.1 points. The implementation code is available at this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2609.37661
PDF
-
Mathematical Transfer in LLMs Follows Reasoning Approach More Than Topic
2026-09-29 13:50:02Sajad Goudarzi, Samaneh Zamanifard, Seyed Amin Seyed Haeri, Moloud Nasiri, Hamed RahimianAbstract
When selecting mathematical training data for LLMs, a natural organizing principle is topic: probability examples for probability targets. An alternative is reasoning approach: worked solutions that share a solution method with the target, even when the mathematical domain differs. We ask which relation produces greater transfer after fine-tuning. We evaluate two counterbalanced $2\times2$ designs: probability and combinatorics crossed with invariant reasoning and double counting (2,000 problems), and number theory and geometry crossed with complement and pigeonhole reasoning (800 problems). In each design, every cell serves as the held-out target in turn: same-approach (SA) sources share the target's method but change the topic, while same-topic (ST) sources share the topic but change the method. Every source appears once in each role, so additive source-quality effects cancel from the equally weighted aggregate contrast. Across five base models and three training seeds per design, SA outperforms ST in all 40 seed-pooled model--target comparisons. Model-level advantages range from 8.2 to 16.2 percentage points in the primary design (mean: 10.8) and from 12.0 to 16.0 in the second design (mean: 14.3); all ten model-level 95% confidence intervals exclude zero. In both designs, ST sources are more similar to targets under embedding and lexical measures, so the SA advantage runs opposite to the measured ordering of statement-level resemblance. These findings identify reasoning approach as a more effective matching criterion than topic for mathematical transfer across the evaluated topic--approach combinations.
Abstract (translated)
URL
https://arxiv.org/abs/2610.00331
PDF
-
Decompose Radicals, Then Reward: Fine-Grained Inspection for Accurate Chinese Text Rendering
2026-09-29 13:37:32Yazhen Xie, Xingsong Ye, Zhineng ChenAbstract
Rendering accurate Chinese text remains challenging for text-to-image models. Existing OCR-based reinforcement-learning rewards compare decoded transcripts with target strings. Such rewards overlook the compositional nature of Chinese writing: an ideograph consists of reusable components arranged through explicit spatial relations, yet OCR evaluates it as an atomic character. Consequently, visually different radical-level errors may receive equally coarse feedback, encouraging glyphs that merely resemble the target instead of faithfully reproducing its internal structure. We employ Ideographic Description Sequences (IDS), which comprise spatial operators and character components, and train an expert IDS recognizer to transcribe rendered Chinese text into this representation. Building on this recognizer, we introduce IDSpect, which deterministically decomposes the target text into IDS tokens and aligns crop-level visual IDS predictions with the target sequence. Globally unique token credit makes this comparison robust to the order of detected text regions. Combined with a whole-character semantic reward, IDSpect supplies fine-grained credit with component and spatial-relation without changing the image generator or adding inference-time cost. Experiments with GRPO post-training of Qwen-Image demonstrate that IDSpect achieves leading structural quality and semantic alignment on LongText and GenTextEval.
Abstract (translated)
URL
https://arxiv.org/abs/2609.37569
PDF
-
E-MoE: Enhanced Mixture-of-Experts for Non-Factorized Diffusion Language Models
2026-09-29 13:18:50Arseny Ivanov, Alexander Kolesov, Alexander Korotin, Ivan Oseledets, Mikhail GoncharovAbstract
Masked diffusion models (MDMs) generate sequences by progressively unmasking several tokens per denoising step, but their reverse process is typically factorized over positions, limiting sample quality in the few-step regime where diffusion's speed advantage over autoregressive decoding matters most. A recent line of work introduces a continuous Gaussian latent, trained as a variational autoencoder, to capture correlations across positions, but such approaches are prone to posterior collapse, where the latent is silently ignored. We propose Enhanced Mixture-of-Experts (E-MoE), which builds the reverse process as a mixture of factorized distributions over a discrete shared latent given by the expert-routing decisions of a Mixture-of-Experts (MoE) backbone, without increasing active parameters over the factorized baseline. Across synthetic multi-modal benchmarks, binarized MNIST, and LM1B, E-MoE improves few-step generation over factorized baselines.
Abstract (translated)
URL
https://arxiv.org/abs/2609.37533
PDF
-
Demistifying Data and Simulator Assumptions in Supervised Causal Discovery
2026-09-29 13:03:22Pingchuan Ma, Rui Ding, Bojun Huang, Shuai WangAbstract
Supervised causal discovery learns to infer causal structure for a new dataset from training datasets paired with structural labels. These training pairs are typically simulated, making the simulator both a source of supervision and a carrier of assumptions about causal graphs, mechanisms, and noise. Understanding the resulting predictions therefore requires examining how these assumptions supplement the information available in observational data, which may be compatible with multiple causal graphs. This paper examines that relationship across representative methods available through June 2026. We organize these methods by prediction target, prediction granularity, encoder, structural decoder, and training regime to relate what each method predicts to how it uses data and simulator-based supervision. Using this framework, we distinguish two questions: whether the target is identifiable under the assumed model class, and whether a trained predictor generalizes beyond its training distribution. Restrictions on mechanisms and noise can make otherwise ambiguous causal directions identifiable, but predictive accuracy under those restrictions does not establish transfer when they change. This distinction motivates evaluation that matches metrics to the identifiable graph target and tests changes in graphs, mechanisms, and noise between training and deployment. Extending such evaluation to real data also requires documenting the external causal evidence and uncertainty behind benchmark reference graphs. Together, these analyses guide method comparison and identify open questions in transfer, test-time adaptation, and uncertainty assessment.
Abstract (translated)
URL
https://arxiv.org/abs/2609.37446
PDF
-
Complexity-Aware Evaluation of LLM Comprehension
2026-09-29 12:42:55Ali Mohammadi Esfahani, Nafiseh Kahani, Samuel A. AjilaAbstract
Large language models (LLMs) are increasingly used for software engineering tasks that require understanding existing source code, including behavior prediction, function explanation, debugging, and code review. However, aggregate benchmark accuracy can conceal how model reliability changes as source code becomes structurally more complex. This paper presents a complexity-aware framework for evaluating LLM code comprehension using cyclomatic complexity, nesting depth, branching factor, and Halstead volume. We evaluate DeepSeek-Coder-V2 and Llama through two complementary tasks: automatic input-output prediction over 300 Python functions and manually assessed semantic comprehension over a balanced subset of 60 functions. The functions are grouped into Low-, Medium-, and High-complexity bands. DeepSeek-Coder-V2 achieves an overall automatic accuracy of 78.33%, compared with 70.33% for Llama. However, accuracy decreases substantially from Low to High complexity, from 93.52% to 52.78% for DeepSeek-Coder-V2 and from 87.04% to 47.22% for Llama. Incorrect predictions are consistently associated with higher values of all four complexity metrics, and correlation and logistic-regression analyses confirm broadly comparable negative associations between structural complexity and correctness. Manual semantic comprehension shows the same degradation pattern, with accuracy decreasing from 100.00% to 75.00% for DeepSeek-Coder-V2 and from 90.00% to 60.00% for Llama. These findings demonstrate that complexity-aware evaluation provides a more diagnostic assessment of LLM code-comprehension reliability than aggregate accuracy alone.
Abstract (translated)
URL
https://arxiv.org/abs/2609.37405
PDF
-
Beyond Prompt Count: How Data Shapes Transfer in On-Policy Distillation
2026-09-29 12:30:10Jiaxuan Wang, Jiafei Lyu, Yuchen Cai, Siye Wu, Pengyuan Wang, Jiashun Liu, Xiang Cheng, Kai Yang, Yangkun Chen, Saiyong Yang, Lan-Zhe GuoAbstract
On-policy distillation (OPD) trains students using teacher feedback on their own sampled responses, yet how prompt choice shapes transfer across teacher-student pairs remains poorly understood. We systematically study prompt quantity, source, and selection across RL- and SFT-continuation pairs and cross-model settings. We find that OPD can be highly prompt-efficient: a few prompts can approach large-pool performance, with four DAPO prompts matching the observed mathematics score of 3,840 DeepMath prompts. However, prompt utility is relational rather than intrinsic: changing only the teacher can reverse the relative effectiveness of mathematics and code prompts. To characterize these transfer differences, we analyze parameter and functional changes across prompt supports and model pairs. Functional alignment with the teacher varies across supports and target tasks; in continuation pairs, teacher-aligned prediction changes can coexist with weak parameter alignment. Continued OPD on effective supports can restore performance after unfavorable transfer. Finally, targeted selection does not consistently outperform uniform random sampling, and filtering out a source that performs poorly alone yields no consistent gain across three paired support draws. Overall, our results distinguish prompt efficiency from prompt interchangeability and show that effective data choice depends on the teacher-student pair and target capability, with random sampling providing a competitive baseline in the studied settings.
Abstract (translated)
URL
https://arxiv.org/abs/2609.37377
PDF
-
Knowing Is Not Choosing: What Explicit Verification Adds Beyond Generative Preference
2026-09-29 12:05:15Yilong Li, Chengpo Yan, Aayan Arish, Suman BanerjeeAbstract
Generating a correct answer does not mean that a language model will select it. We separate factual recall into three steps: generating a correct candidate, ranking the available candidates, and selecting the final answer. Pre-generation readouts predict factual recall and which questions sampling will cover across three model families, but say little about whether an available correct answer will ultimately be selected. Explicit verification with $P(\mathrm{True})$ improves within-question ranking over mean log-likelihood in Gemma, Qwen3, and Llama, with AUROC gains of $0.08$--$0.12$. In a prospectively defined Gemma cohort, verification raises plurality accuracy by about $5$ points, and still gains about $2$ points over chat-template likelihood, a stronger generative baseline. The advantage is strongest for relations with common-answer priors and depends on access to the entity; masking the entity removes the ranking advantage in larger Qwen models. Finally, the measured benefit depends on how correctness is defined: recall-oriented reference matching can credit option lists favored by likelihood and substantially understate the improvement seen under human semantic judgments. Prior work shows that models can carry latent factual knowledge and judge candidate answers; we show that these capabilities do not collapse into a single notion of ``knowing,'' and trace where information is gained, lost, or mismeasured between availability, ranking, and final choice.
Abstract (translated)
URL
https://arxiv.org/abs/2609.33142
PDF
-
VISTA: Value-Informed Event Appraisal for Multimodal Emotion Conflict
2026-09-29 12:01:15Jiale Dai, Liuxian Ma, Xiaoke Niu, Wenjing Zhang, Huiying Zhao, Zhaoxiang Liu, Shiguo Lian, Guojie SongAbstract
Conflicting emotional cues can be individually valid: a subdued voice may reflect a blocked goal while a smile satisfies a social obligation. Their interpretation depends on what the event means to the person. We introduce VISTA (Value-Informed Semantic Trust Arbitration), a learned seven-field appraisal interface that conditions modality arbitration on concerns, event relations, and expression conditions while retaining a joint-evidence residual. A log-odds decomposition separates emotion expectation from cue diagnosticity, motivating an interface that lets appraisal change how evidence is interpreted. With a shared Qwen2.5-Omni-7B backbone and matched training examples and steps, VISTA reaches 64.5% conflict accuracy on CA-MER, improving on modality gating by 2.5 percentage points on conflict and 0.2 on consistency. Shuffling appraisal across scenes or removing its decision connection reduces this benefit. A common frozen-backbone probe reaches 0.600 macro CCC for appraisal readout, compared with 0.505 for emotion-only fine-tuning. Evaluations across five benchmarks connect recognition under increasing conflict with appraisal readout and downstream decision use. Together, the analyses and experiments support scene-specific appraisal as an intermediate representation that helps interpret conflicting emotional evidence.
Abstract (translated)
URL
https://arxiv.org/abs/2609.37324
PDF
