Computation and Language · Cross-listed from cs.SE
Learning from Execution: Self-Evolving Memory for Private-Library Code Generation
Mofei Li, Taozhi Chen, Guowei Yang et al. (+1 more)
Large
Language Models (LLMs) have achieved strong performance on general code generation, but their effectiveness drops sharply in enterprise settings where software development relies on internal private libraries absent from public pre-training corpora. Existing
Retrieval-Augmented Generation (RAG) methods provide a training-free solution by retrieving static API documentation, but our analysis shows that documentation mainly helps models identify what APIs to use and remains insufficient for teaching how to use them correctly. Even with oracle API-document
retrieval, LLMs still make recurring errors at the API, cross-API, and task levels, including API misuse or
hallucination, flawed API composition, and incorrect solution strategies. To address this limitation, we propose MEMCoder, a training-free self-evolving
memory framework for private-library code generation. MEMCoder augments existing RAG pipelines with a Multi-level Evolving
Memory that continuously accumulates and reuses execution-derived Usage Guidelines at the API, cross-API, and task levels. During generation, MEMCoder retrieves both static API documentation and relevant historical
memories to guide code generation; after execution, it analyzes feedback to refine
memory through a closed loop of generation, execution, reflection, and update. Extensive experiments on NdonnxEval and NumbaEval show that MEMCoder consistently enhances different RAG backbones across LLMs of different scales, yielding an average absolute pass@1 improvement of 18.41 percentage points. Moreover, MEMCoder outperforms existing self-evolving
memory methods and validates the effectiveness of organizing execution feedback into multi-level usage
memories.
Read the full abstract ▾
All authors Mofei Li, Taozhi Chen, Guowei Yang, Jia Li
Computation and Language
Rewarding the Scientific Process: Process-Level Reward Modeling for Agentic Data Analysis
Zhisong Qiu, Shuofei Qiao, Kewei Xu et al. (+4 more)
Process Reward Models (PRMs) have achieved remarkable success in augmenting the
reasoning capabilities of Large
Language Models (LLMs) within static domains such as mathematics. However, their potential in dynamic data analysis tasks remains underexplored. In this work, we first present a empirical study revealing that general-domain PRMs struggle to supervise data analysis
agents. Specifically, they fail to detect silent errors, logical flaws that yield incorrect results without triggering interpreter exceptions, and erroneously penalize exploratory actions, mistaking necessary trial-and-error exploration for grounding failures. To bridge this gap, we introduce DataPRM, a novel environment-aware generative process reward model that (1) can serve as an active verifier, autonomously interacting with the environment to probe intermediate execution states and uncover silent errors, and (2) employs a reflection-aware ternary reward strategy that distinguishes between correctable grounding errors and irrecoverable mistakes. We design a scalable pipeline to construct over 8K high-quality training instances for DataPRM via diversity-driven trajectory generation and knowledge-augmented step-level annotation. Experimental results demonstrate that DataPRM improves downstream policy LLMs by 7.21% on ScienceAgentBench and 11.28% on DABStep using Best-of-N inference. Notably, with only 4B parameters, DataPRM outperforms strong baselines, and exhibits robust generalizability across diverse Test-Time Scaling strategies. Furthermore, integrating DataPRM into Reinforcement Learning yields substantial gains over outcome-reward baselines, achieving 78.73% on DABench and 64.84% on TableBench, validating the effectiveness of process reward supervision. Code is available at https://github.com/zjunlp/DataMind.
Read the full abstract ▾
All authors Zhisong Qiu, Shuofei Qiao, Kewei Xu, Yuqi Zhu, Lun Du, Ningyu Zhang, Huajun Chen
Computation and Language
Seeing Is No Longer Believing: Frontier Image Generation Models, Synthetic Visual Evidence, and Real-World Risk
Shuai Wu, Xue Li, Zhijun Wang et al. (+4 more)
Image generation systems can produce plausible photographs, readable documents, and consistent depictions of people and places. When these artifacts are presented as records of real events, they can influence decisions in news, finance, identity verification, medicine, and law. This narrative review examines selected public model documentation, incident reports, research, and governance sources available through 1 October 2026, with English and Chinese community material providing illustrative context. We distinguish vendor capability claims, documented incidents, experimental findings, and prospective harm pathways. The analysis connects realism, text rendering, reference consistency, editing, grounding, and production cost to the conditions under which synthetic images acquire evidentiary authority. Historical incidents illustrate these pathways; they do not establish misuse rates for current models. We compare provider restrictions, provenance systems, watermarking, platform labeling, and policy obligations, and explain how their functions differ from independent verification of a depicted event or transaction. The framework links artifact types and decision contexts to the functions of available controls. High-stakes decisions call for authenticated source records, corroboration through trusted channels, and proportionate review before action.
Read the full abstract ▾
All authors Shuai Wu, Xue Li, Zhijun Wang, Bolun Liu, Weilin Cai, Zihao Su, Ran Wang
Computation and Language
MultiDx: A Multi-Source Knowledge Integration Framework towards Diagnostic Reasoning
Yimin Deng, Zhenxi Lin, Yejing Wang et al. (+9 more)
Diagnostic prediction and clinical
reasoning are critical tasks in healthcare applications. While Large
Language Models (LLMs) have shown strong capabilities in commonsense
reasoning, they still struggle with diagnostic
reasoning due to limited domain knowledge. Existing approaches often rely on internal model knowledge or static knowledge bases, resulting in knowledge insufficiency and limited adaptability, which hinder their capacity to perform diagnostic
reasoning. Moreover, these methods focus solely on the accuracy of final predictions, overlooking
alignment with standard clinical
reasoning trajectories. To this end, we propose MultiDx, a two-stage diagnostic
reasoning framework that performs differential diagnosis by analyzing evidence collected from multiple knowledge sources. Specifically, it first generates suspected diagnoses and
reasoning paths by leveraging knowledge from web search, SOAP-formatted case, and clinical case database. Then it integrates multi-perspective evidence through matching, voting, and differential diagnosis to generate the final prediction.~Extensive experiments on two public
benchmarks demonstrate the effectiveness of our approach.
Read the full abstract ▾
All authors Yimin Deng, Zhenxi Lin, Yejing Wang, Guoshuai Zhao, Pengyue Jia, Zichuan Fu, Derong Xu, Yefeng Zheng, Xiangyu Zhao, Li Zhu, Xian Wu, Xueming Qian
Computation and Language
MemeScouts@LT-EDI 2026: Asking the Right Questions – Prompted Weak Supervision for Meme Hate Speech Detection
Ivo Bueno, Lea Hirlimann and Enkelejda Kasneci
Detecting hate
speech in memes is challenging due to their
multimodal nature and subtle, culturally grounded cues such as sarcasm and context. While recent
vision-
language models (VLMs) enable joint
reasoning over text and images, end-to-end prompting can be brittle, as a single prediction must resolve target, stance, implicitness, and irony. These challenges are amplified in
multilingual settings. We propose a prompted weak supervision (PWS) approach that decomposes meme understanding into targeted, question-based labeling functions with constrained answer options for homophobia and transphobia detection in the LT-EDI 2026 shared task. Using a quantized Qwen3-VLM to extract features by answering targeted questions, our method outperforms direct VLM classification, with substantial gains for Chinese and Hindi, ranking 1st in English, 2nd in Chinese, and 3rd in Hindi. Iterative refinement via error-driven LF expansion and feature
pruning reduces redundancy and improves generalization. Our results highlight the effectiveness of prompted weak supervision for
multilingual multimodal hate
speech detection.
Read the full abstract ▾
Computation and Language
AdapTime: Enabling Adaptive Temporal Reasoning in Large Language Models
Yimin Deng, Yejing Wang, Zhenxi Lin et al. (+8 more)
Large
language models have demonstrated strong
reasoning capabilities in general knowledge question answering. However, their ability to handle temporal information remains limited. To address this limitation, existing approaches often involve external tools or manual verification and are tailored to specific scenarios, leading to poor generalizability. Moreover, these methods apply a fixed pipeline to all questions, overlooking the fact that different types of temporal questions require distinct
reasoning strategies, which leads to unnecessary processing for simple cases and inadequate
reasoning for complex ones. To this end, we propose AdapTime, an adaptive temporal
reasoning method that dynamically executes
reasoning steps based on the input context. Specifically, it involves three temporal
reasoning actions: reformulate, rewrite and review, with an LLM planner guiding the
reasoning process. AdapTime integrates seamlessly with state-of-the-art LLMs and significantly enhances their temporal
reasoning capabilities without relying on external support. Extensive experiments demonstrate the effectiveness of our approach.
Read the full abstract ▾
All authors Yimin Deng, Yejing Wang, Zhenxi Lin, Zichuan Fu, Guoshuai Zhao, Derong Xu, Yefeng Zheng, Xiangyu Zhao, Xian Wu, Li Zhu, Xueming Qian
Computation and Language
Psychologically-Grounded Graph Modeling for Interpretable Depression Detection
Rishitej Reddy Vyalla, Kritarth Prasad, Avinash Anand et al. (+4 more)
Automatic depression detection from conversational interactions holds significant promise for scalable screening but remains hindered by severe data scarcity and a lack of clinical interpretability. Existing approaches typically rely on black-box deep learning architectures that struggle to model the subtle, temporal
evolution of depressive symptoms or account for participant-specific heterogeneity. In this work, we propose PsyGAT (Psychological Graph
Attention Network), a psychologically grounded framework that models conversational sessions as dynamic temporal graphs. We introduce Psychological Expression Units (PEUs) to explicitly encode utterance-level clinical evidence, structuring the session graph to capture transitions in psychological states rather than mere semantic dependencies. To address the critical class imbalance in depression
datasets, we employ clinically approved persona-based data augmentation, enable robust model learning. Additionally, we integrate session-level personality context directly into the graph structure to disentangle trait-based behavior from acute depressive symptoms. PsyGAT achieves state-of-the-art performance, surpassing both strong graph-based baselines and closed-source LLMs like GPT-5, achieving 89.99 and 71.37 Macro F1 scores in DAIC-WoZ and E-DAIC, respectively. We further introduce Causal-PsyGAT, an interpretability module that identifies symptom triggers. Experiments show a 20% improvement in MRR for identifying causal indicators, effectively bridging the gap between depression monitoring and clinical explainability. The full augmented
dataset is publicly available at https://doi.org/10.6084/m9.figshare.31801921.
Read the full abstract ▾
All authors Rishitej Reddy Vyalla, Kritarth Prasad, Avinash Anand, Erik Cambria, Shaoxiong Ji, Faten S. Alamri, Zhengkui Wang
Computation and Language
IRIS: Interleaved Reinforcement with Incremental Staged Curriculum for Cross-Lingual Mathematical Reasoning
Navya Gupta, Rishitej Reddy Vyalla, Avinash Anand et al. (+8 more)
Curriculum learning helps
language models tackle complex
reasoning by gradually increasing task difficulty. However, it often fails to generate consistent step-by-step
reasoning, especially in
multilingual and low-resource settings where cross-lingual transfer from English to Indian
languages remains limited. We propose IRIS: Interleaved Reinforcement with Incremental Staged Curriculum, a two-axis framework that combines Supervised Fine-Tuning on progressively harder problems (vertical axis) with Reverse Curriculum Reinforcement Learning to reduce reliance on step-by-step guidance (horizontal axis). We design a composite reward combining correctness, step-wise
alignment, continuity, and numeric incentives, optimized via Group Relative Policy Optimization (GRPO). We release CL-Math, a
dataset of 29k problems with step-level annotations in English, Hindi, and Marathi. Across standard
benchmarks and curated
multilingual test sets, IRIS consistently improves performance, with strong results on math
reasoning tasks and substantial gains in low-resource and bilingual settings, alongside modest improvements in high-resource
languages.
Read the full abstract ▾
All authors Navya Gupta, Rishitej Reddy Vyalla, Avinash Anand, Chhavi Kirtani, Erik Cambria, Zhengchen Zhang, Zhengkui Wang, Timothy Liu, Aik Beng Ng, Simon See, Rajiv Ratn Shah
Computation and Language
Factual and Edit-Sensitive Graph-to-Sequence Generation via Graph-Aware Adaptive Noising
Aditya Hemant Shahane, Anuj Kumar Sirohi, Tanmoy Chakraborty et al. (+2 more)
Fine-tuned autoregressive models for graph-to-sequence generation (G2S) often struggle with factual grounding and edit sensitivity. To tackle these issues, we propose a non-autoregressive
diffusion framework that generates text by iterative refinement conditioned on an input graph, named as
Diffusion Language Model for Graphs (DLM4G). By aligning graph components (entities/relations) with their corresponding sequence
tokens, DLM4G employs an adaptive noising strategy. The proposed strategy uses per-
token denoising error as a signal to adaptively modulate noise on entity and relation
tokens, improving preservation of graph structure and enabling localized updates under graph edits. Evaluated on three
datasets, DLM4G consistently outperforms competitive G2S
diffusion baselines trained on identical splits across both surface-form and
embedding-based metrics. DLM4G further exceeds fine-tuned autoregressive baselines up to 12x larger (e.g., T5-Large) and is competitive with zero-shot LLM transfer baselines up to 127x larger. Relative to the strongest fine-tuned PLM baseline, DLM4G improves factual grounding (FGT@0.5) by +5.16% and edit sensitivity (ESR) by +7.9%; compared to the best
diffusion baseline, it yields gains of +3.75% in FGT@0.5 and +23.6% in ESR. We additionally demonstrate applicability beyond textual graphs through experiments on molecule captioning, indicating the method's generality for scientific G2S generation.
Read the full abstract ▾
All authors Aditya Hemant Shahane, Anuj Kumar Sirohi, Tanmoy Chakraborty, Prathosh A P, Sandeep Kumar
Computation and Language
BiMol-Diff: A Unified Diffusion Framework for Molecular Generation and Captioning
Aditya Hemant Shahane, Anuj Kumar Sirohi, Devansh Arora et al. (+3 more)
Bridging molecular structures and natural
language is essential for controllable design. Autoregressive models struggle with long-range dependencies, while standard
diffusion processes apply uniform corruption across positions, which can distort structurally informative
tokens. We present BiMol-Diff, a unified
diffusion framework for the paired tasks of text-conditioned molecule generation and molecule captioning. Our key component is a
token-aware noise schedule that assigns position-dependent corruption based on
token recovery difficulty, preserving harder-to-recover substructures during the forward process. On ChEBI-20 and M3-20M, BiMol-Diff improves molecule reconstruction with a 15.4% relative gain in Exact Match and achieves strong captioning results, attaining best BLEU and BERTScore among compared baselines. These results indicate
token-aware noising improves fidelity in molecular structure-
language modelling.
Read the full abstract ▾
All authors Aditya Hemant Shahane, Anuj Kumar Sirohi, Devansh Arora, Nitin Kumar, Prathosh A P, Sandeep Kumar
Computation and Language · Cross-listed from cs.AI
Information-Geometric First-Passage Monitoring of Distributional Stability in Stochastic Systems
Hikmat Karimov and Rahid Zahid Alekberli
Runtime monitoring of
stochastic systems must distinguish nominal distributional relaxation from regime departure while controlling repeated-test false alarms under explicit validity assumptions. This paper links relative-
entropy dissipation, information geometry, and sequential inference in a bounded first-passage monitoring architecture. For reversible Fokker–Planck dynamics, relative
entropy to an invariant density is non-increasing; under exogenous forcing, its derivative decomposes into nominal dissipation and an information-space forcing term. The runtime layer uses Gaussian window surrogates, nominal-relative covariance shrinkage, a coordinate-consistent relative precision diagnostic, and randomized conformal ranks aggregated by a mixture power-martingale process. Analytical Ornstein–Uhlenbeck validation gives zero positive nominal Kullback–Leibler increments, forcing-identity residuals below 3.31 x 10^-6, and coordinate-invariance errors at numerical roundoff. On NSL-KDD, the monitor yields 0/100 alarms on internal nominal streams but 63/100 on official test-normal streams; post-change detection is 99.0% for seen and 98.53% for test-only attack types with median one-window delay. On UNSW-NB15, internal-null alarms are 0/100, whereas official test-normal alarms rise to 90/100; post-change detection is 81.33%, with 18.67% pre-change alarms. In these evaluations, calibration transport emerges as a major deployment constraint. No universal
benchmark superiority, causal inference, or physical-work interpretation is claimed.
Read the full abstract ▾
Computation and Language · Cross-listed from cs.CR
Jailbreaking Frontier Foundation Models Through Intention Deception
Xinhe Wang, Katia Sycara and Yaqi Xie
Large (
vision-)
language models exhibit remarkable capability but remain highly susceptible to jailbreaking. Existing safety training approaches aim to have the model learn a refusal boundary between safe and unsafe, based on the user's intent. It has been found that this binary training regime often leads to brittleness, since the user intent cannot reliably be evaluated, especially if the attacker obfuscates their intent, and also makes the system seem unhelpful. In response, frontier models, such as GPT-5, have shifted from refusal-based safeguards to safe completion, that aims to maximize helpfulness while obeying safety constraints. However, safe completion could be exploited when a user pretends their intention is benign. Specifically, this intent inversion would be effective in multi-turn conversation, where the attacker has multiple opportunities to reinforce their deceptively benign intent. In this work, we introduce a novel multi-turn jailbreaking method that exploits this vulnerability. Our approach gradually builds conversational trust by simulating benign-seeming intentions and by exploiting the consistency property of the model, ultimately guiding the target model toward harmful, detailed outputs. Most crucially, our approach also uncovered an additional class of model vulnerability that we call para-jailbreaking that has been unnoticed up to now. Para-jailbreaking describes the situation where the model may not reveal harmful direct reply to the attack query, however the information that it reveals is nevertheless harmful. Our contributions are threefold. First, it achieves high success rates against frontier models including GPT-5-thinking and Claude-Sonnet-4.5. Second, our approach revealed and addressed para-jailbreaking harmful output. Third, experiments on
multimodal VLM models showed that our approach outperformed state-of-the-art models.
Read the full abstract ▾
Computation and Language
The Pragmatic Persona: Discovering LLM Persona through Bridging Inference
Jisoo Yang, Jongwon Ryu, Minuk Ma et al. (+2 more)
Large
Language Models (LLMs) reveal inherent and distinctive personas through dialogue. However, most existing persona discovery approaches rely on surface-level lexical or stylistic cues, treating dialogue as a flat sequence of
tokens and failing to capture the deeper discourse-level structures that sustain persona consistency. To address this limitation, we propose a novel analytical framework that interprets LLM dialogue through bridging inference – implicit conceptual relations that connect utterances via shared world knowledge and discourse coherence. By modeling these relations as structured knowledge graphs, our approach captures latent semantic links that govern how LLMs organize meaning across turns, enabling persona discovery at the level of discourse coherence rather than surface realizations. Experimental results across multiple
reasoning backbones and target LLMs, ranging from small-scale models to 80B-parameter systems, demonstrate that bridging-inference graphs yield significantly stronger semantic coherence and more stable persona identification than frequency or style-based baselines. These results show that persona traits are consistently encoded in the structural organization of discourse rather than isolated lexical patterns. This work presents a systematic framework for probing, extracting, and visualizing latent LLM personas through the lens of Cognitive Discourse Theory, bridging computational linguistics, cognitive semantics, and persona
reasoning in large
language models. Codes are available at https://github.com/JiSoo-Yang/Persona_Bridging.git
Read the full abstract ▾
All authors Jisoo Yang, Jongwon Ryu, Minuk Ma, Trung X. Pham, Junyeong Kim
Computation and Language · Cross-listed from cs.AI
An Information-Geometric Framework for Stability Analysis of Large Language Models under Entropic Stress
Hikmat Karimov and Rahid Zahid Alekberli
As large
language models (LLMs) are increasingly deployed in high-stakes and operational settings, evaluation strategies based solely on aggregate accuracy are often insucient to characterize system reliability. This study proposes a thermodynamic inspired modeling framework for analyzing the stability of LLM outputs under conditions of uncertainty and perturbation. The framework introduces a composite stability score that integrates task utility,
entropy as a measure of external uncertainty, and two internal structural proxies: internal integration and aligned reective capacity. Rather than interpreting these quantities as physical variables, the formulation is intended as an interpretable abstraction that captures how internal structure may modulate the impact of disorder on model behavior. Using the IST-20 benchmarking protocol and associated metadata, we analyze 80 modelscenario observations across four contemporary LLMs. The proposed formulation consistently yields higher stability scores than a reduced utilityentropy baseline, with a mean improvement of 0.0299 (95% CI: 0.02470.0351). The observed gain is more pronounced under higher
entropy conditions, suggesting that the framework captures a form of nonlinear attenuation of uncertainty. We do not claim a fundamental physical law or a complete theory of machine ethics. Instead, the contribution of this work is a compact and interpretable modeling perspective that connects uncertainty, performance, and internal structure within a unied evaluation lens. The framework is intended to complement existing benchmarking approaches and to support ongoing discussions in AI safety, reliability, and governance.
Read the full abstract ▾
Computation and Language
How Sensitive Are Safety Benchmarks to Judge Configuration Choices?
Xinran Zhang
Safety
benchmarks such as HarmBench rely on LLM judges to classify model responses as harmful or safe, yet the judge configuration, namely the combination of judge model and judge
prompt, is typically treated as a fixed implementation detail. We show this assumption is problematic. Using a 2 x 2 x 3 factorial design, we construct 12 judge
prompt variants along two axes, evaluation structure and instruction framing, and apply them using a single judge model, Claude Sonnet 4-6, producing 28,812 judgments over six target models and 400 HarmBench behaviors. We find that
prompt wording alone, holding the judge model fixed, shifts measured harmful-response rates by up to 24.2 percentage points, with even within-condition surface rewording causing swings of up to 20.1 percentage points. Model safety rankings are moderately unstable, with mean Kendall tau = 0.89, and category-level sensitivity ranges from 39.6 percentage points for copyright to 0 percentage points for harassment. A supplementary multi-judge experiment using three judge models shows that judge-model choice adds further variance. Our results demonstrate that judge
prompt wording is a substantial, previously under-examined source of measurement variance in safety benchmarking.
Read the full abstract ▾
Computation and Language
PeeriScope: A Multi-Faceted Framework for Evaluating Peer Review Quality
Sajad Ebrahimi, Soroush Sadeghian, Ali Ghorbanpour et al. (+6 more)
The increasing scale and variability of peer review in scholarly venues has created an urgent need for systematic, interpretable, and extensible tools to assess review quality. We present PeeriScope, a modular platform that integrates structured features, rubric-guided large
language model assessments, and supervised prediction to evaluate peer review quality along multiple dimensions. Designed for openness and integration, PeeriScope provides both a public interface and a documented API, supporting practical deployment and research extensibility. The demonstration illustrates its use for reviewer self-assessment, editorial triage, and large-scale
auditing, and it enables the continued development of quality evaluation methods within scientific peer review. PeeriScope is available both as a live demo at https://app.reviewer.ly/app/peeriscope and via API services at https://github.com/Reviewerly-Inc/Peeriscope.
Read the full abstract ▾
All authors Sajad Ebrahimi, Soroush Sadeghian, Ali Ghorbanpour, Negar Arabzadeh, Sara Salamat, Seyed Mohammad Hosseini, Hai Son Le, Mahdi Bashari, Ebrahim Bagheri
Computation and Language
Distilling Self-Consistency into Verbal Confidence: A Pre-Registered Negative Result and Post-Hoc Rescue on Gemma 3 4B
Jon-Paul Cacioli
Small instruct-tuned LLMs produce degenerate verbal confidence under minimal elicitation: ceiling rates above 95%, near-chance Type-2 AUROC, and Invalid validity profiles. We test whether confidence-conditioned supervised fine-tuning (CSFT) with self-consistency-derived targets can close the gap between internal information and verbal readout. A pre-registered Phase 0 protocol on Gemma 3 4B-it with a modal filter restricting training to items with correct modal answers produced a negative result: AUROC2 dropped from 0.554 to 0.509 due to label-
entropy collapse in the training targets. An exploratory rescue removed the filter, training on all 2,000 calibration items. This produced a binary verbal correctness discriminator with AUROC2 = 0.774 on held-out TriviaQA, compressing a 10-sample self-consistency signal (AUROC2 = 0.999) into a single-pass readout exceeding logit
entropy (0.701). The shuffled-target control showed no improvement (0.501). On MMLU, accuracy improved from 54.2% to 77.4% with the shuffled model at baseline (56.1%), supporting a target-dependent interpretation. The result is exploratory, binary rather than continuously calibrated, and observed at a single scale. It identifies two design lessons: confidence training requires label
entropy, and correct targets regularise output format.
Read the full abstract ▾
Computation and Language
Improving Robustness of Tabular Retrieval via Representational Stability
Kushal Raj Bhandari, Adarsh Singh, Jianxi Gao et al. (+2 more)
Transformer-based table
retrieval systems flatten structured tables into
token sequences, making
retrieval sensitive to the choice of serialization even when table semantics remain unchanged. We show that semantically equivalent serializations, such as $\texttt{csv}$, $\texttt{tsv}$, $\texttt{html}$, $\texttt{markdown}$, and $\texttt{ddl}$, can produce substantially different
embeddings and
retrieval results across multiple
benchmarks and retriever families. To address this instability, we treat serialization
embedding as noisy views of a shared semantic signal and use its centroid as a canonical target representation. We show that centroid averaging suppresses format-specific variation and can recover the semantic content common to different serializations when format-induced shifts differ across tables. Empirically, centroid representations outrank individual formats in aggregate pairwise comparisons across $\texttt{MPNet}$, $\texttt{BGE-M3}$, $\texttt{ReasonIR}$, and $\texttt{SPLADE}$. We further introduce a lightweight residual bottleneck adapter on top of a frozen encoder that maps single-serialization
embeddings towards centroid targets while preserving variance and enforcing covariance regularization. The adapter improves robustness for several dense retrievers, though gains are model-dependent and weaker for sparse lexical
retrieval. These results identify serialization sensitivity as a major source of
retrieval variance and show the promise of post hoc
geometric correction for serialization-invariant table
retrieval.
Read the full abstract ▾
All authors Kushal Raj Bhandari, Adarsh Singh, Jianxi Gao, Soham Dan, Vivek Gupta
Computation and Language · Cross-listed from cs.LG
AgenticCache: Cache-Driven Asynchronous Planning for Embodied AI Agents
Hojoon Kim, Yuheng Wu and Thierry Tambe
Embodied AI
agents increasingly rely on large
language models (LLMs) for planning, yet per-step LLM calls impose severe latency and cost. In this paper, we show that
embodied tasks exhibit strong plan locality, where the next plan is largely predictable from the current one. Building on this, we introduce AgenticCache, a planning framework that reuses cached plans to avoid per-step LLM calls. In AgenticCache, each
agent queries a runtime cache of frequent plan transitions, while a background Cache Updater asynchronously calls the LLM to validate and refine cached entries. Across four multi-
agent embodied benchmarks, AgenticCache improves task success rate by 22% on average across 12 configurations (4
benchmarks x 3 models), reduces simulation latency by 65%, and lowers
token usage by 50%. Cache-based plan reuse thus offers a practical path to low-latency, low-cost
embodied agents. Code is available at https://github.com/hojoonleokim/MLSys26_AgenticCache.
Read the full abstract ▾
Computation and Language · Cross-listed from cs.AI
AgentPulse: A Continuous Multi-Signal Framework for Evaluating AI Agents in Deployment
Yuxuan Gao, Megan Wang and Yi Ling Yu
Static
benchmarks measure what AI
agents can do at a fixed point in time but not how they are adopted, maintained, or experienced in deployment. We introduce AgentPulse, a continuous evaluation framework scoring 50
agents across 10 workload categories along four factors (
Benchmark Performance, Adoption Signals, Community Sentiment, and Ecosystem Health) aggregated from 18 real-time signals across GitHub, package registries, IDE marketplaces, social platforms, and
benchmark leaderboards. Three analyses ground the framework. The four factors capture largely complementary information (n=50; $\rho_{\max}=0.61$ for Adoption-Ecosystem, all others $|\rho| \leq 0.37$). A circularity-controlled test (n=35) shows the
Benchmark+Sentiment sub-composite, which contains no GitHub-derived signals, predicts external adoption proxies it does not aggregate: GitHub
stars ($\rho_s=0.52$, $p<0.01$) and Stack Overflow question volume ($\rho_s=0.49$, $p<0.01$), with VS Code installs ($\rho_s=0.44$, $p<0.05$) reported as illustrative given that only 11 of 35
agents have non-zero installs. On the n=11 subset with published SWE-bench scores, composite and
benchmark-only rankings are nearly uncorrelated ($\rho_s=0.25$; 9 of 11
agents shift by at least 2 ranks), driven by a strong negative Adoption-Capability correlation among closed-source high-capability
agents within this subset. This is precisely why we rest the framework's validity claim on the broader n=35 test rather than the SWE-bench overlap. AgentPulse surfaces deployment signal absent from
benchmarks; it is a methodology, not a ground-truth ranking. The framework, all collected signals, scoring outputs, and evaluation harness are released under CC BY 4.0.
Read the full abstract ▾
Computation and Language · Cross-listed from cs.CV
DeepTaxon: An Interpretable Retrieval-Augmented Multimodal Framework for Unified Species Identification and Discovery
Jiawei Wang, Ming Lei, Yaning Yang et al. (+8 more)
Identifying species in biology among tens of thousands of visually similar taxa while discovering unknown species in open-world environments remains a fundamental challenge in biodiversity research. Current methods treat identification and discovery as separate problems, with classification models assuming closed sets and discovery relying on threshold-based rejection. Here we present DeepTaxon, a
retrieval-augmented
multimodal framework that unifies species identification and discovery through interpretable
reasoning over retrieved visual evidence. Given a query image, DeepTaxon retrieves the top-$k$ candidate species with $n$ exemplar images each from a
retrieval index and performs chain-of-thought comparative
reasoning. Critically, we redefine discovery as an explicit,
retrieval-based decision problem rather than an implicit parametric
memory problem. A sample is novel if and only if the
retrieval index lacks sufficient evidence for identification, so each
retrieval naturally yields a classification or discovery label without manual annotation, thereby providing automatic supervision for both tasks. We train the framework via supervised fine-tuning on synthetic
retrieval-augmented data, followed by reinforcement learning on hard samples, converting high-recall
retrieval into high-precision decisions that scale to massive taxonomic vocabularies. Extensive experiments on a large-scale in-distribution
benchmark and six out-of-distribution
datasets demonstrate consistent improvements in both identification and discovery. Ablation studies further reveal effective test-time scaling with candidate count $k$ and exemplar count $n$, strong zero-shot transfer to unseen domains, and consistent performance across
retrieval encoders, establishing an interpretable solution for biodiversity research.
Read the full abstract ▾
All authors Jiawei Wang, Ming Lei, Yaning Yang, Xinyan Lin, Yuquan Le, Qiwei Ma, Zhiwei Xu, Zheqi Lv, Yuchen Ang, Zhe Quan, Tat-Seng Chua
Computation and Language
From Skill Text to Skill Structure: The Scheduling-Structural-Logical Representation for Agent Skills
Qiliang Liang, Hansi Wang, Zhong Liang et al. (+1 more)
Large
language model (LLM)
agents increasingly rely on reusable skills: capability packages that combine instructions, control flow, constraints, and tool calls. In current
agent systems, however, skills are still represented by text-heavy artifacts, mainly SKILL{.}md-style documents whose machine-usable evidence remains embedded largely in natural-
language descriptions. As a result, skill-centered
agent systems face a representation problem: both managing skill collections and using skills during
agent execution require
reasoning over invocation interfaces, execution structure, and concrete side effects, but these signals are often entangled in a single textual surface. An explicit representation of skill knowledge may therefore help make these artifacts easier for machines to acquire and leverage. Drawing on
Memory Organization Packets, Script Theory, and Conceptual Dependency from Schank and Abelson's classical work on cognitive linguistic representation, we introduce what is, to our knowledge, the first structured representation for
agent skill artifacts that disentangles skill-level scheduling signals, scene-level execution structure, and logic-level action/resource-use evidence: the Scheduling-Structural-Logical (SSL) representation. We instantiate SSL with an LLM-based normalizer and evaluate SSL-derived representations in two tasks, Skill Discovery and Risk Assessment. The experiment shows that SSL significantly outperforms the text-only baselines: in Skill Discovery, MRR@50 improves from 0.649 to 0.729; in Risk Assessment, macro F1 improves from 0.409 to 0.509. These findings suggest that an explicit, source-grounded structure can make
agent skills easier to search and review, positioning SSL as a practical step toward more inspectable, reusable, and operationally actionable skill representations, rather than a finished standard or end-to-end skill-management mechanism.
Read the full abstract ▾
All authors Qiliang Liang, Hansi Wang, Zhong Liang, Yang Liu
Computation and Language
Stabilizing Efficient Reasoning with Step-Level Advantage Selection
Han Wang, Xiaodong Yu, Jialian Wu et al. (+4 more)
Large
language models (LLMs) achieve strong
reasoning performance by allocating substantial computation at inference time, often generating long and verbose
reasoning traces. While recent work on efficient
reasoning reduces this overhead through length-based rewards or
pruning, many approaches are post-trained under a much shorter context window than base-model training, a factor whose effect has not been systematically isolated. We first show that short-context post-training alone, using standard GRPO without any length-aware objective, already induces substantial
reasoning compression-but at the cost of increasingly unstable training dynamics and accuracy degradation. To address this, we propose Step-level Advantage Selection (SAS), which operates at the
reasoning-step level and assigns a zero advantage to low-confidence steps in correct rollouts and to high-confidence steps in verifier-failed rollouts, where failures often arise from truncation or verifier issues rather than incorrect
reasoning. Across diverse mathematical and general
reasoning benchmarks, SAS improves average Pass@1 accuracy by 0.86 points over the strongest length-aware baseline while reducing average
reasoning length by 16.3%, yielding a better accuracy-efficiency trade-off.
Read the full abstract ▾
All authors Han Wang, Xiaodong Yu, Jialian Wu, Jiang Liu, Ximeng Sun, Mohit Bansal, Zicheng Liu
Computation and Language · Cross-listed from cs.LG
When to Commit? Towards Variable-Size Self-Contained Blocks for Discrete Diffusion Language Models
Danny Wang, Ruihong Qiu and Zi Huang
Discrete
diffusion language models (dLLMs) enable parallel
token updates with bidirectional
attention, yet practical generation typically adopts blockwise semi-autoregressive decoding. This switch creates a training-inference mismatch: training denoises with full-sequence context, while inference commits
tokens within a bounded block without future context. Therefore, decoding with fixed-size or heuristic-based blocks can lead to premature
token commitments, as decisions are made without full access to future context that could alter those choices. Motivated by this, we propose self-containedness as a principled criterion for block commitment. A block is self-contained if its predictions remain consistent with Future-Aware (FA) or without No-Future (NF) access to future context, reframing block boundary selection as a test of self-containedness rather than a heuristic choice. Based on this principle, we introduce Variable-size Self-contained Blocks (VSB) for dLLMs. VSB scores and selects block boundaries using the divergence between
token-level predictive distributions under NF and FA conditioning, which quantifies how predictions would change if future context were revealed. We provide theoretical justification linking self-containedness to predictive consistency, and extensive experiments validate VSB's efficacy over fixed-size and heuristic blockwise decoding.
Read the full abstract ▾
Computation and Language
EPM-RL: Reinforcement Learning for On-Premise Product Mapping in E-Commerce
Minhyeong Yu and Wonduk Seo
Product mapping, the task of deciding whether two e-commerce listings refer to the same product, is a core problem for price monitoring and channel visibility. In real marketplaces, however, sellers frequently inject promotional keywords, platform-specific tags, and bundle descriptions into titles, causing the same product to appear under many different names. Recent LLM-based and multi-
agent frameworks improve robustness and interpretability on such hard cases, but they often rely on expensive external APIs, repeated
retrieval, and complex inference-time orchestration, making large-scale deployment costly and difficult in
privacy-sensitive enterprise settings. To address these issues, we present EPM-RL, a reinforcement-learning-based framework for building an accurate and efficient on-premise e-commerce product mapping model. Our central idea is to distill high-cost agentic
reasoning into a trainable in-house model. Starting from a curated set of product pairs with LLM-generated rationales and human verification, we first perform parameter-efficient fine-tuning (PEFT) on a small student model using structured
reasoning outputs. We then further optimize the model with Reinforcement Learning (RL) using an
agent-based reward that jointly evaluates output-format compliance, label correctness,
reasoning–preference scores from specially designed judge models. Preliminary results show that EPM-RL consistently improves over PEFT-only training and offers a stronger quality–cost trade-off than commercial API-based baselines, while enabling private deployment and lower operational cost. These findings suggest that reinforcement learning can turn product mapping from a high-latency agentic pipeline into a scalable, inspectable, and production-ready in-house system.
Read the full abstract ▾
Computation and Language · Cross-listed from cs.AI
Representational Curvature Modulates Behavioral Uncertainty in Large Language Models
Jack King, Evelina Fedorenko and Eghbal A. Hosseini
In autoregressive large
language models (LLMs), temporal straightening offers an account of how the next-
token prediction objective shapes representations. Models learn to progressively straighten the representational trajectory of input sequences across layers, potentially facilitating next-
token prediction via linear extrapolation. However, a direct link between this trajectory and
token-level behavior has been missing. We provide such a link by relating contextual
curvature-a
geometric measure of how sharply the representational trajectory bends over recent context-to next-
token entropy. Across two models (GPT-2 XL and Pythia-2.8B), contextual
curvature is correlated with
entropy, and this relationship emerges during training. Perturbation experiments reveal selective dependence: manipulating
curvature through trajectory-aligned interventions reliably modulates
entropy, while geometrically misaligned perturbations have no effect. Finally, regularizing representations to be straighter during training modestly reduces
token-level
entropy without degrading validation loss. These results identify trajectory
curvature as a task-aligned representational feature that influences behavioral uncertainty in LLMs.
Read the full abstract ▾
Computation and Language
Propagation Structure-Semantic Transfer Learning for Robust Fake News Detection
Mengyang Chen, Lingwei Wei, Han Cao et al. (+3 more)
Fake news generally refers to false information that is spread deliberately to deceive people, which has detrimental social effects. Existing fake news detection methods primarily learn the semantic features from news content or integrate structural features from propagation. However, in practical scenarios, due to the semantic ambiguity of informal
language and unreliable user interactive behaviors on social media, there are inherent semantic and structural noises in news content and propagation. Although some recent works consider the effect of irrelevant user interactions in a hybrid-modeling way, they still suffer from the mutual interference between structural noise and semantic noise, leading to limited performance for robust detection. To alleviate this issue, this paper proposes a novel Propagation Structure-Semantic Transfer Learning framework (PSS-TL) for robust fake news detection under a teacher-student architecture. Specifically, we design dual teacher models to learn semantics knowledge and structure knowledge from noisy news content and propagation structure independently. Besides, we design a Multi-channel Knowledge
Distillation (MKD) loss to enable the student model to acquire specialized knowledge from the teacher models, thereby avoiding mutual interference. Extensive experiments on two real-world
datasets validate the effectiveness and robustness of our method.
Read the full abstract ▾
All authors Mengyang Chen, Lingwei Wei, Han Cao, Wei Zhou, Zhou Yan, Songlin Hu
Computation and Language
Quantum Knowledge Graph: Modeling Context-Dependent Triplet Validity
Yao Wang, Zixu Geng and Jun Yan
Knowledge graphs (KGs) are increasingly used to support large lan guage model (LLM)
reasoning, but standard triplet-based KGs treat each relation as globally valid. In many settings, whether a relation should count as evidence depends on the context. We therefore formulate triplet validity as a triplet-specific function of context and refer to this formulation as a
Quantum Knowledge Graph (QKG). We instantiate QKG in medicine using a diabetes-centered PrimeKG subgraph, whose 68,651 context-sensitive relations are further annotated with patient-group-specific constraints. We evaluate it in a reasoner–validator pipeline for medical question answering on a KG-grounded subset of MedReason containing 2,788 questions. With Haiku-4.5 as both the Reasoner and the Validator, KG-backed validation significantly improves over a no-validator baseline ($+0.61$ pp), and QKG with context matching yields the largest gain, outperforming both KG validation without context matching ($+0.79$ pp) and the no-validator baseline ($+1.40$ pp; paired McNemar, all $p<0.05$). Under a stronger validator (Qwen-3.6-Plus), the raw QKG gain over the no-validator baseline grows from $+1.40$ pp to $+5.96$ pp; the context-matching gap is non-significant ($p=0.73$) on the raw set but becomes borderline significant ($p=0.05$) after adjustment for knowledge leakage and suspicious questions, consistent with a
benchmark-gold ceiling rather than a QKG limitation. Taken together, the results support the view that the value of a KG in LLM-based clinical
reasoning lies not merely in storing medically related facts, but in representing whether those facts are applicable to the specific patient context. For reproducibility and further research, we release the curated QKG
datasets and source code.\footnote{https://github.com/HKAI-Sci/QKG}
Read the full abstract ▾
Computation and Language
KOMBO: Korean Character Representations Based on the Combination Rules of Subcharacters
SungHo Kim, Juhyeong Park, Yeachan Kim et al. (+1 more)
The Korean writing system, Hangeul, has a unique character representation rigidly following the invention principles recorded in Hunminjeongeum.\footnote{Hunminjeongeum is a book published in 1446 that describes the principles of invention and usage of Hangeul, devised by King Sejong \cite{Hunminjeongeum_Guide}.} However, existing pre-trained
language models (PLMs) for Korean have overlooked these principles. In this paper, we introduce a novel framework for Korean PLMs called KOMBO, which firstly brings the invention principles of Hangeul to represent character. Our proposed method, KOMBO, exhibits notable experimental proficiency across diverse NLP tasks. In particular, our method outperforms the state-of-the-art Korean PLM by an average of 2.11% in five Korean natural
language understanding tasks. Furthermore, extensive experiments demonstrate that our proposed method is suitable for comprehending the linguistic features of the Korean
language. Consequently, we shed light on the superiority of using subcharacters over the typical subword-based approach for Korean PLMs. Our code is available at: [https://github.com/SungHo3268/KOMBO](https://github.com/SungHo3268/KOMBO).
Read the full abstract ▾
All authors SungHo Kim, Juhyeong Park, Yeachan Kim, SangKeun Lee
Computation and Language
TSAssistant: A Human-in-the-Loop Agentic Framework for Automated Target Safety Assessment
Xiaochen Zheng, Zhiwen Jiang, David Tokar et al. (+5 more)
Target Safety Assessment (TSA) requires systematic integration of genetic, transcriptomic, target homology, pharmacological, and clinical data to evaluate potential safety liabilities of therapeutic targets. This process is labor-intensive and expert-dependent, posing challenges in scalability and reproducibility. We present TSAssistant, a human-in-the-loop multi-
agent framework that decomposes TSA report generation into a workflow of specialized subagents: Research Subagents that each ground and cite a single TSA domain, and Synthesis Subagents that integrate findings across domains. Subagents retrieve and synthesize evidence from curated biomedical sources through standardized tool interfaces and produce individually citable, evidence-grounded sections, with behavior shaped by a hierarchical instruction architecture that separates coordination logic from domain expertise and user intent. To complement these soft constraints, programmatic execution hooks and persistent
memory stores enforce hard constraints across the workflow, while an interactive refinement loop allows experts to review and revise individual sections with full conversational context preserved across iterations. Rather than a single holistic comparison, we decompose report quality into reproducibility, evidential grounding, task-level accuracy, and controllability under expert oversight, finding high reproducibility and grounding, substantial agreement with the human reference, and net-positive expert-driven refinement.
Read the full abstract ▾
All authors Xiaochen Zheng, Zhiwen Jiang, David Tokar, Yexiang Cheng, Alvaro Serra, Melanie Guerard, Klas Hatje, Tatyana Doktorova