Peer Revenue Publish or profit

Papers

Fresh from arXiv, typeset properly. Highlighted words trade on the market.

Computer Science Computation and Language

cs.CL Watch Computation and Language

Thursday 8 October

Computation and Language

HarnessSQL: Harness-Native Training for SQL Agents in Realistic Database Environments

Haolin Yang, Jipeng Zhang, Zhiqiang Qian et al. (+4 more)

Text-to-SQL models are commonly trained to map questions directly to static queries, whereas real-world database agents operate through stateful, multi-turn interaction with live databases – inspecting schemas, executing probe queries, diagnosing errors, and revising hypotheses. This creates a critical train-deploy mismatch, as the execution harness that mediates this interaction is introduced only at inference time. To bridge this gap, we propose HarnessSQL, a harness-native post-training framework that preserves the full interaction structure throughout both supervised fine-tuning and reinforcement learning. HarnessSQL builds isolated, executable database environments paired with hidden execution oracles, rolls out teachers directly inside the target SQL harness, and retains only verified trajectories for full-sequence SFT, followed by execution-reward RL. Across Spider 2.0-SQLite, HarnessSQL dramatically boosts the execution accuracy of compact models, raising Qwen3-8B from 15.5% to 45.2% and Qwen3-14B from 22.2% to 54.8%, while transferring effectively to out-of-distribution interactive benchmarks such as BIRD-Interact and LiveSQLBench. Our findings demonstrate that training database agents directly within their execution harness is essential for mastering complex, long-horizon database workflows.
Read the full abstract

All authors Haolin Yang, Jipeng Zhang, Zhiqiang Qian, Jian Xie, Shuaishuai Gong, Yike Guo, Sirui Han

Computation and Language

EgoVoice: Proactive Spoken Assistance from Egocentric Multimodal Streams

Heeseung Kim

Wearable augmented reality (AR) assistants are moving toward continuous real-world interaction, where they perceive the user's activity through first-person video and audio and provide timely spoken guidance without being explicitly asked. While proactive video assistants, spoken dialog systems, and egocentric task understanding have each advanced rapidly, existing systems do not address the joint problem of deciding when to speak and what to say from continuous first-person streams. We introduce EgoVoice, a framework for training and evaluating proactive egocentric spoken assistants. From HoloAssist video recordings of real human instructors, we construct clean audio streams through source separation and speech resynthesis, and convert each video session into a format where the model must decide at each moment whether to remain silent or provide spoken guidance. We fine-tune an omni-modal LLM with our data, and further improve its proactive intervention behavior with direct preference optimization. Experiments across closed and open-source models show that existing systems rarely produce well-timed, meaningful proactive interventions, while EgoVoice yields clear improvements in intervention timing, content relevance, and human preference over the zero-shot backbone.
Read the full abstract

Computation and Language · Cross-listed from cs.IR

NativeScope: Relation-Localized Retrieval over Native Topology with a Correct Anchor

Long Wang

Dense retrieval usually ranks text chunks by their semantic similarity to a question. This ignores structure that many data systems already store, including section membership, session boundaries, and native order. We propose NativeScope, a scope-then-rank method for queries with a known anchor and relation. It represents a query as q -> (A, r, B). The anchor A and relation r select native units through belonging, before, or after operators, and the target term B ranks only chunks that overlap the selected scope. An internal variant, NS-FullQ, ranks the same candidates with the full question. We evaluate both methods on 200 controlled document and memory records derived from QASPER and LongMemEval under a 1,024-token budget. NativeScope attains native-unit recall of 89.28 percent for documents and 72.50 percent for memories, improving over instance-wide Dense RAG by 42.75 and 22.00 percentage points. NS-FullQ reaches 87.78 percent and 68.50 percent; its differences from NativeScope are inconclusive, locating the primary gain in relational scoping rather than the shorter ranking query. With automatic Top-1 anchors, memory recall falls to 35.50 percent. NativeScope is therefore effective when anchor coordinates and native relations are reliable, but hard scoping inherits errors from the localization interface.
Read the full abstract

Computation and Language

TokenRouter: Efficient Serving System for Token-Level LLM Routing

Tianyu Fu, Tengxuan Liu, Ruoxi Wang et al. (+4 more)

Large language model (LLM) routing distributes inference work across different models, advancing the cost-quality Pareto frontier of LLM serving. While coarse-grained routing at the session or query level has been widely adopted in production systems, recent algorithmic work shows that fine-grained token-level routing can yield substantial efficiency and quality gains. However, efficiently serving token-level routed inference poses significant challenges to existing systems. Built on single-LLM assumptions, current systems suffer from severe step desynchronization and frequent batch admission delays under token-level routing, and they also impose high implementation complexity on developers. To address these challenges, we design TokenRouter, an efficient and developer-friendly serving system for token-level routed LLM inference. TokenRouter follows the principle of request-centric programming, model-centric execution: developers describe routing logic from the perspective of a single request, while the runtime launches a subserver for each LLM and dispatches requests asynchronously. Each subserver employs a delayed-batching scheduler, whose optimal hyperparameters are derived from a mathematical throughput model of the system. Across diverse routing algorithms, workloads, and model pairs, TokenRouter achieves 2.01-64.15x higher decoding throughput than existing systems, substantially advancing the serving efficiency of token-level LLM routing. Our code is available at https://github.com/thu-nics/TokenRouter.
Read the full abstract

All authors Tianyu Fu, Tengxuan Liu, Ruoxi Wang, Yixin Dong, Yi Ge, Yichen You, Yu Wang

Computation and Language

Language Models as AI Research World Models

Zijun Wang, Zewen Liu, Minhua Lin et al. (+10 more)

AI research agents automate the cycle of proposing, implementing, and evaluating experiments, opening a path toward recursive self-improvement. Yet their ability to propose experiments outpaces their capacity to execute them in real environments, making outcome prediction a key capability for sustained self-improvement under limited experimental budgets. We investigate language models as Research World Models (RWMs), which predict the outcomes of candidate interventions across research environments. Our evaluation draws on over 2,600 experimental records from nine research environments spanning pretraining, post-training, and inference, representing more than 171,000 H100 GPU-hours of experimentation. Research knowledge acquired from real experimental experience improves RWM predictions of unseen interventions within the same environment (Spearman +0.27), and can be reused across environments. For example, using only pretraining experience from OLMo3, Marin, and Nanochat, an RWM reduces selection regret in the Qwen3 environment by 78% compared with zero-experience setting. These benefits extend to multi-round Autoresearch under a fixed selection budget: RWMs with in-env and cross-env research knowledge increase the best gain achieved by 15.8% and 11.6%, respectively. Ablations across 13 language models used as RWMs show that adding research knowledge can improve intervention ranking more than changing models or increasing reasoning effort alone. These findings support language models as RWMs and motivate accumulating experimental data for future RWM training.
Read the full abstract

All authors Zijun Wang, Zewen Liu, Minhua Lin, Zhaotian Weng, Zhan Shi, Bing He, Yisi Sang, Dakuo Wang, Benoit Dumoulin, Wei Jin, Yuyin Zhou, Cihang Xie, Hanqing Lu

Computation and Language · Cross-listed from cs.SD

DiffuPlex: Accelerating Full-Duplex Spoken Dialog Models via Rolling Masked Diffusion

Heeseung Kim

Recent full-duplex spoken dialog models enable simultaneous listening and speaking, but fine-grained models still advance their backbone autoregressively at every interaction frame. We introduce DiffuPlex, a rolling masked diffusion framework that reduces this sequential computation by predicting multiple future user and assistant frames in a single backbone wake. DiffuPlex consumes only a confident prefix of each predicted future while interaction continues at the original frame rate. As user speech arrives, it checks the corresponding user predictions and, when the interaction diverges, preserves already played assistant content while revising only the unplayed future. We consider two inference policies over the same predictor: DiffuPlex-LISTEN consumes multiple future frames when they predict assistant silence, whereas DiffuPlex-SPEAK can also consume predicted assistant speech. Across full-duplex interaction and spoken-language evaluations, DiffuPlex substantially reduces sequential backbone computation while largely preserving interaction behavior and general capability. DiffuPlex-LISTEN and DiffuPlex-SPEAK achieve $1.46\times$ and $1.59\times$ deployment-path wall-clock speedups and $1.61\times$ and $1.80\times$ Core LM speedups, with all measured backbone invocations completing within the 80ms interaction interval. Human evaluation shows that LISTEN preserves speech naturalness and conversational quality, while SPEAK retains conversational quality with some degradation in speech naturalness.
Read the full abstract

Computation and Language

SciTBERT: A family of chronologically consistent language models for scientific and technological language processing

Thomas Gebhart and Russell J. Funk

Pre-trained transformer models are increasingly being used to study scientific and technological progress. Encoders tuned to paper or patent text outperform general-purpose models on downstream classification, regression, and proximity tasks within science and technology. However, the applicability of these models for studying time-dependent or archival properties of science, technology, and their interface is limited due to lookahead and domain biases inherent to these pre-trained models. These limitations arise from training on corpora with unconstrained chronological and text source distributions. We introduce SciTBERT: a family of chronologically consistent BERT-derived language models trained on text from scientific papers, patents, and high-quality educational web text with training data cutoff dates spanning each year between 2013 and 2025. We also post-train these models in a chronologically-consistent manner using paper and patent citations, creating SciTBERT-CI model family. We find that these models generally outperform predecessor domain-specific encoder models even when training data is limited by early year restrictions in the corpus. To further investigate the extent to which this class of models can learn representations that bridge science and technology, we introduce the PatRepEval benchmark, a suite of patent-related text embedding tasks at the science-technology interface. Performance in a variety of classification, regression, and retrieval tasks spanning papers and patents highlights the importance of aligning encoder model representations with the domain distributions of their downstream tasks, and chronologically consistent encoders can match or exceed models trained without temporal constraints.
Read the full abstract

Computation and Language · Cross-listed from eess.AS

SteerablePlex: Can We Steer Full-Duplex Models?

Haolong Zheng, Maike Züfle, Dominik Macháček et al. (+6 more)

Full-duplex speech models can listen and speak simultaneously, enabling natural interaction, but become increasingly difficult to control as the conversation history grows. When used as user simulators, this lack of control can cause them to deviate from prescribed scenarios and produce unreliable evaluation outcomes. We introduce SimIF-Bench (Simulator Instruction-Following Benchmark), which evaluates whether a conversational model stays within a prescribed scenario and completes multiple goals in the required order. The benchmark reveals that current open-source full-duplex models struggle to follow such constraints. We then introduce a Group Reward-Decoupled Normalization Policy Optimization (GDPO)-based training recipe that enables a full-duplex model to follow textual instructions during an ongoing conversation while maintaining its turn-taking ability. By connecting the resulting SteerablePlex to an asynchronous backend language model that monitors the conversation and provides instructions when needed, we build a more controllable full-duplex user simulator that follows multi-stage constraints more reliably than existing open-source models and GPT-Realtime.
Read the full abstract

All authors Haolong Zheng, Maike Züfle, Dominik Macháček, Peter Polák, Xulin Fan, Xavier Sumba, Siyin Wang, Ondřej Klejch, Mark Hasegawa-Johnson

Computation and Language · Cross-listed from cs.LG

When KL Regularization Misfires in Group Policy Optimization

Fei Ding

Why does removing reference-policy KL regularization sometimes improve group policy optimization? This motivates studying how reference-policy information should enter group-relative updates. We analyze seven potential failure modes in the interactions between KL and rewards: residual KL updates after reward clipping, after gradient cancellation, and in groups with identical rewards; KL growth with response length and an imbalance in its relative contribution; KL concentration on a small number of tokens; and sampling noise when k1 is incorporated into rewards. We propose Zero-Sum Calibrated Policy Optimization (ZCPO), which uses relative drift measured by conditional KL to calibrate within-group reward coefficients and integrates them into the base surrogate. Mathematical reasoning experiments and ablations support this design's effectiveness in our settings.
Read the full abstract

Computation and Language

Language-Specific Effects of Tokenizer Choice in Multilingual Language Models

Clara Meister, Gül Sena Alt{\i}ntaş and Antoine Bosselut

Tokenizer choice affects multilingual language modeling, but vocabulary capacity is finite and vocabulary size is often constrained: improving representation for some languages often comes at the expense of others. We therefore ask whether tokenizer choice matters equally across languages, a question that the current literature leave unanswered. To this end, we train 123 language models spanning 54 tokenizers. In the main comparison, architecture, training corpus, training-token budget, and optimization are held fixed, so the models differ only in their tokenizer. We find that tokenizer choice matters more for languages with less language-model training data: across the 54 tokenizers, the standard deviation of a language's bits-per-byte (BPB) increases as its model training-data share decreases (Spearman rho = -0.52 over the 31 trained languages and -0.69 over the 28 written with word boundaries). Leaving a language out of tokenizer training raises its BPB in every language we study, and the penalty tends to be larger for languages with less language-model training data. Giving lower-resource languages a larger share of tokenizer-training data, however, does not unconditionally help those languages: both equal weighting and an allocation inverting the shares with respect to the language model training data increase their BPB, particularly when language-model training repeats data. Finally, which intrinsic tokenizer properties are associated with better BPB differs across languages, providing further evidence that what makes a good tokenizer depends on the language. We find that the metrics quantifying these properties can be successfully used to predict downstream models' pairwise BPB rankings, suggesting a practical strategy for screening tokenizer candidates before training language models.
Read the full abstract

Computation and Language

Rehearse Everything, Remember Nothing: Attic-KV Rehearses What Will Be Read

Zhiyun Shi

Many key-value (KV) caches are compressed before anyone knows what will be asked of them: a document cached for retrieval, a prompt prefix shared across requests, the memory of a long conversation. The prevailing approach scores KV entries by rehearsal: the model rereads the context and keeps the entries it attends to, assuming that the more completely a cache rehearses its context, the better it remembers it. We show that under tight budgets this assumption backfires: rehearse everything, remember nothing. At a 3% keep ratio, rereading the whole context keeps 31.5 of 96.5 points on RULER, and on LongBench's natural-text tasks it falls below methods that rehearse nothing at all. The cause is that a cache keeps what it rehearses: rereading spreads the budget across the whole context, so the answer's own entries survive at little more than chance. Like a student before an exam, a cache remembers more by testing itself than by rereading. Two principles follow: rehearse what will be read, and rehearse as much as there is. We instantiate them as Attic-KV (Attic for short), a training-free rehearsal in which the model quizzes itself with question-answer pairs that quote the context, alongside anchor tokens in a content-adaptive amount. Changing only the rehearsal lifts three hosts that score it in three different ways: Attic alone is the best training-free method in all eight settings we test on RULER and LongBench's natural-text tasks, and plugged into the gradient-based KVgrad and the trained RestoreKV+, it raises them by up to 17.1 and 28.1 points. Its advantage grows as the budget shrinks, reaching 41.9 points over full rereading at a 3% keep ratio, and on long documents it compresses faster than rereading them.
Read the full abstract

Computation and Language

A persistent accuracy ceiling in automated verbal deception detection

Riccardo Loconte, Jonas Festor, Zane Fatjanova et al. (+2 more)

Automated methods have been proposed to overcome the limitations of human verbal deception detection, but evidence remains fragmented across disciplines. We systematically reviewed 25 years of research (289 reports, 6,136 classification models) and meta-analyzed 3,653 models nested within 97 datasets. Pooled accuracy was 74.4% (95% CI: 71.2%-77.4%) with substantial heterogeneity. Accuracy was driven by methodological quality (ground truth, data source, class balance, evaluation procedure) more than by model complexity: the adoption of embeddings and large language models has not translated into improved predictive performance. Only 12.46% of reports used data with verifiable ground-truth, and only 23.96% of models were evaluated on independent data. The pooled accuracy aligns with meta-analyses of manual approaches, suggesting a ceiling of 70-75%, unlikely to be lifted by current research conventions.
Read the full abstract

All authors Riccardo Loconte, Jonas Festor, Zane Fatjanova, Mariam Bolkvadze, Bennett Kleinberg

Computation and Language

All Verdicts are Not Equal: Rethinking LLM Judge Reliability

Vineet Kumar, Darshita Rathore and Anindya Moitra

LLM-as-a-Judge is the standard paradigm for NLP evaluation, yet its systemic reliability remains poorly understood despite being widely treated as a deterministic ground truth. We present a comprehensive reliability audit, stresstesting six frontier models across four benchmarks, five prompt formats, two presentation orders, three sampling temperatures, and ten repetitions per condition. Our empirical analysis reveals severe vulnerabilities: verdicts change across identical replications at temperature zero, position-order swaps flip the majority of verdicts on challenging tasks, and the most deterministic judge achieves perfect consistency by trivially repeating incorrect verdicts, agreeing with ground truth only 51% of the time. To formalize these multi-faceted failure modes, we introduce the trustworthy verdict rate (T ), a unified metric capturing the joint probability that an evaluation is reproducible, order-invariant, and accurate. UsingT , we derive a theoretical upper bound on accuracy imposed by position bias and show that reliability is item-specific rather than modellevel. Finally, we demonstrate that shifting from pairwise win-rate to holistic rubric scoring improves trustworthiness more than any single-format prompting intervention, offering an actionable framework for robust NLP evaluation.
Read the full abstract

Computation and Language

ILM: An AI-Powered Storytelling Educational Tool

Suhaila Mohammed, Abdelaziz Serour and Allison Lahnala

Digital technologies have made Islamic narratives more accessible, but existing platforms provide limited support for structured learning and comprehension of these stories, particularly in Arabic and multilingual settings. We present ILM, an interactive educational platform for Stories of the Prophets that combines Arabic natural language processing, structured knowledge representation, and retrieval-based question generation. Admin-approved Arabic narratives are processed by a Knowledge Graph (KG) Constructor Engine that identifies entities and narrative relationships and stores them as structured knowledge, enabling learners to explore stories through a visual story map and answer entity- and relation-based questions generated from the KG. Separately, a multilingual retrieval pipeline retrieves relevant passages from the original narratives to generate multiple-choice and open-ended comprehension questions. For open-ended questions, an LLM-as-a-Judge evaluates learners' answers against the retrieved passages and reference answers to determine correctness. The platform also incorporates Quranic content as a separate enrichment layer, allowing selected narratives to be supplemented with source-supported information. By combining structured knowledge with passage-based retrieval, ILM supports narrative exploration, comprehension, and assessment across Arabic and multilingual content. The system demonstrates the feasibility of combining structured knowledge representation and retrieval-based generation to support interactive learning of Islamic narratives. A demo is available at anonymous.4open.science/r/mml-5FCF.
Read the full abstract

Computation and Language · Cross-listed from cs.AI

When Should Agents Think? Adaptive Reasoning via Cross-Turn Estimation

Yiruo Cheng, Shen Huang, Xiaoshuai Song et al. (+5 more)

Large language model (LLM)-based agents have demonstrated strong capabilities on complex tasks. They typically perform reasoning before each action throughout an interaction trajectory. However, reasoning may not be necessary at every turn, as reasoning produced earlier can continue to support subsequent actions. A key challenge is therefore to determine when existing reasoning remains sufficient and when a new reasoning step is needed, without relying on costly generation-based verification. We find that decreases in the likelihood of subsequent reference actions after removing additional reasoning closely track whether those actions remain recoverable given earlier reasoning, providing an effective and lightweight signal for estimating cross-turn action support. Based on this observation, we propose Reasoning Adaptation through Cross-Turn Estimation (RACE), a training approach for adaptive agent reasoning. RACE introduces a Likelihood-Guided Progressive Reasoning Cover Detection (LoGiC) procedure that progressively identifies reasoning turns whose removal has limited impact on the current and subsequent reference actions. The resulting removal signals are incorporated into both supervised fine-tuning and agentic reinforcement learning, enabling the policy to learn when to reason and when to act directly. Extensive experiments on four representative agent benchmarks show that RACE substantially reduces reasoning cost while maintaining or improving task performance.
Read the full abstract

All authors Yiruo Cheng, Shen Huang, Xiaoshuai Song, Jiejun Tan, Guanting Dong, Pengjun Xie, Ji-Rong Wen, Zhicheng Dou

Computation and Language

Natural Language to First-Order Logic LLM-based Autoformalization

Andrea Brunello, Cristian Curaba, Luca Geatti et al. (+3 more)

Large Language Models (LLMs) have renewed interest in autoformalization. Yet, when First-Order Logic (FOL) is considered as the target formalism, the field still lacks a unified task formulation and a systematic survey. This paper addresses this gap: we first provide a principled definition for the FOL-autoformalization task by distinguishing Ontology Extraction from Logical Translation, showing how their conflation obscures (cross-study) evaluation; we review existing datasets, evaluation metrics, and LLM-based methods, including fine-tuning, prompting, and verification-based refinement; we identify open challenges in benchmarking, semantic evaluation, ontology-aware methods, and end-to-end applications.
Read the full abstract

All authors Andrea Brunello, Cristian Curaba, Luca Geatti, Michele Mignani, Angelo Montanari, Nicola Saccomanno

Computation and Language · Cross-listed from cs.AI

InterviewPlayground: A Validated Simulation Environment for Evaluating AI Interviewers

Jonathan Ivey, Aimee Liang, Arthur Y.S. Wang et al. (+3 more)

Increasingly, AI interviewers are being developed to elicit open-ended responses in applications like market research, public polling, preference elicitation, and social science research. However, evaluating AI interviewers is challenging because they function in extended, multi-turn interactions where they must adapt to participant behaviors. To address this need, we develop InterviewPlayground, a simulation environment for evaluating AI interviewers using simulated study participants whose behaviors are grounded in social theory. Simulated studies in InterviewPlayground produce an InterviewReportCard, which assesses the performance of AI interviewers using a suite of validated measures. To test whether our simulation-based evaluations predict performance with human participants, we conduct 15 real qualitative studies with five AI interviewers, three interview topics, and 450 human participants and compare them to simulated studies in InterviewPlayground. We find that AI interviewer performance in InterviewPlayground predicts performance in human studies with an average Pearson correlation of 0.86 across 12 measures, and the simulated interactions from InterviewPlayground reproduce key findings from behavioral analysis of AI interviewers in the human studies. Together, these findings support the validity of InterviewPlayground in assessing AI interviewer performance and examining potential failure modes. Our work contributes a simulation environment for AI interviewers supported with empirical validation, and more broadly, a roadmap for future work to develop validated, simulation-based evaluations of conversational AI systems.
Read the full abstract

All authors Jonathan Ivey, Aimee Liang, Arthur Y.S. Wang, Madeline Mandell, Ziang Xiao, Anjalie Field

Computation and Language

Examining Social Attribution in LLM Reasoning: A Theory-Guided Probing Methodology

Zhaoxin Yu, Qingchao Kong, Dajun Zeng et al. (+1 more)

Large language models (LLMs) are increasingly deployed in sociotechnical systems where social attribution, the reasoning process attributing external events to the causes and reasons of agents' social behaviors, plays a critical role. These processes involve judgments of social cause, responsibility, and blame/credit to agents. Although attributional models are well-studied in social psychology and cognition through Attribution Theory, social attribution remains underexplored in AI, particularly LLM social reasoning. This paper provides the first systematic exploration of LLM social attribution. Our work focuses on responsibility and blame attributions, examining current LLMs' judgments and their underlying internal mechanisms. Guided by attribution theory, we construct a social attribution benchmark consisting of a Vignette subset based on classic scenarios from attribution theory research and a Reality subset based on real-world social narratives, yielding 7,639 responsibility/blame judgment questions. On this basis, we evaluate 32 representative LLMs and 5 basic non-LLM baselines. To further explore the internal mechanisms underlying the LLM judgment process, we develop a probing-based methodology to investigate the latent-space representations of 5 key attribution dimensions and the consistency of their influences on LLM judgments compared to those in human social attribution. Our research findings reveal that current LLMs exhibit measurable but incomplete agreement with human responsibility and blame judgments, and meanwhile, this agreement is positively correlated with model size. Some attribution dimensions are systematically decodable from specific positions in LLM hidden states, and their influences on the final judgment are consistent with those indicated by human Attribution Theory. The dataset and associated code are available at https://github.com/Yuzhaoxin946/SAB-Bench.
Read the full abstract

All authors Zhaoxin Yu, Qingchao Kong, Dajun Zeng, Wenji Mao

Computation and Language · Cross-listed from cs.LG

Agentic-TTT: Training test-time policy for test-time training

Jiahao Lu and Mohan Kankanhalli

Test-time training (TTT) adapts an LLM's parameters using signals derived from test inputs, and can make striking improvements in pre-specified settings such as IMO competitions or designated open problems. By turning deployment experience into parameter updates, TTT provides a direct mechanism for model-level self-improvement. Yet TTT is not universally beneficial: each TTT algorithm works in different settings, and applying an ill-suited method could waste test-time compute or even damage model performance. Therefore, such parameter-level self-improvement requires agency: the model must decide when TTT is warranted, which algorithm to invoke, and whether an existing skill can be reused. To fill this gap, we introduce Agentic-TTT, which learns a test-time policy to govern those decisions. Agentic-TTT turns TTT procedures into callable tools, treats accumulated skills as an evolving deployment environment, and trains its policy using the observed utility gains from its decisions. On our benchmark, Agentic-TTT nearly doubles the utility over the backbone model, learns to trade off utility against compute, and generalizes to domains unseen during training. Together, these results point toward autonomous self-improvement: models that can decide how to learn from their own deployment experience.
Read the full abstract

Computation and Language · Cross-listed from cs.CV

DataVista: Diagnosing Multimodal LLMs on Data Video Understanding

Yupeng Xie, Zhenyang Wang, Jiayi Zhu et al. (+4 more)

Data video is a media form that integrates data visualization with video narrative, widely adopted in news reporting and business analysis. Compared with general video understanding, data video understanding places greater emphasis on accurately reading data from animated charts, integrating evidence across charts and time, and understanding how narrative organization and visual design communicate information. Yet existing benchmarks target either general videos or static charts, and data video understanding has not been systematically evaluated. We present DataVista, the first benchmark for data video understanding, containing 961 real-world data videos and 6,775 evaluation questions organized under a three-level progressive capability framework (data perception, temporal reasoning, narrative understanding) with 10 fine-grained question types across five topic domains. Systematic evaluation of 19 mainstream MLLMs shows that the best-performing model, Gemini-3.1-Pro, achieves 70.0% overall accuracy, still far below human expert performance, with models performing worst on Causal Reasoning and Narrative Structure. Increasing frame counts and adding subtitles mainly benefit data perception and temporal reasoning, with limited gains in narrative understanding. Further analysis of model responses identifies typical failure modes in chart reading, evidence judgment, and instruction understanding. The benchmark is available at https://github.com/HKUSTDial/DataVista.
Read the full abstract

All authors Yupeng Xie, Zhenyang Wang, Jiayi Zhu, Yinghao Tang, Zhouan Shen, Yiyu Chen, Yuyu Luo

Computation and Language

Specialized Decision Models vs. General-Purpose LLMs: Benchmarking Jev Across Knowledge, Reasoning, and Multilingual Tasks

Xing Li, Qingcheng Chang, Jinzhong Ning et al. (+5 more)

Jev is a "System One" model that returns a choice among given options instead of generating text. We study how such a specialized decision model compares with general-purpose large language models (LLMs). We evaluate Jev on 13 multiple-choice benchmarks covering knowledge, reasoning, and multilingual understanding, and compare it with 19 LLMs in three tiers: frontier, representative, and small. Jev is competitive with frontier LLMs on knowledge and commonsense benchmarks and obtains the best score on MMLU-Redux and ARC-Challenge. Outside mathematics, it also outperforms most representative LLMs and all small LLMs. However, it falls behind on mathematical word problems: on MathQA, it is 17.7 points below the frontier median and scores lower than all 19 LLMs. These results indicate that a specialized decision model can match general-purpose LLMs on decisions that rely mainly on knowledge, but not on decisions that require multi-step calculation.
Read the full abstract

All authors Xing Li, Qingcheng Chang, Jinzhong Ning, Changfeng Xu, Shenlong Zhang, Yijia Zhang, Ling Luo, Hongfei Lin

Computation and Language · Cross-listed from cs.AI

MindFlow: Mind Supernet Powered Thinking Flows for Research Idea Innovation

Mengdi Liu, Wenjue Chen, Wenyue Chen et al. (+10 more)

Research idea innovation is a fundamental engine of scientific progress, yet it remains difficult to generate and evaluate in a scalable and controllable way. This challenge lies in its inherently open-ended and multi-objective nature, where ideas should balance novelty, plausibility and feasibility. While recent LLM-based approaches have made progress through carefully designed prompts or agent pipelines, they are constrained by predefined, static ideation workflows. To address this limitation, we propose MindFlow, a framework that explicitly formulates ideation as a graph-structured Flow in Mind, which is composed of modular thinking operators and modeled by a probabilistic mind supernet. Given a research topic, a controller dynamically samples thinking flows to generate candidate ideas. This open-ended problem is optimized using a tournament-based relative ranking, enabling the controller to progressively favor higher-quality thinking flows. We further introduce an evaluation protocol that jointly assesses problem finding and problem solving, going beyond title- or abstractonly judgments. Across diverse topics, MindFlow shows its superiority as an explicit, controllable and optimizable research idea innovator.
Read the full abstract

All authors Mengdi Liu, Wenjue Chen, Wenyue Chen, Cheng Yang, Fanqi Kong, Zhangyang Gao, Xiaoxue Cheng, Yiheng Li, Yujian Yuan, Keliang Li, Hong Chang, Shiguang Shan, Chenglin Wu

Computation and Language

MiMo-V2.6: Scaling Reinforcement Learning Towards Self-Improvement

Xiaomi LLM-Core Team: Zongming Qiao, Ziyue Hua, Zirui Ou et al. (+145 more)

Reinforcement learning (RL) is the central training paradigm for advancing large foundation models towards self-improvement. This report introduces the MiMo-V2.6 series, an omni-modal family that pushes the frontier of model intelligence by scaling RL compute. Prior to RL, we conduct mid-training on a broad multimodal corpus to provide ample exploration space, and build a solid infrastructure on the pretrained hybrid-SWA architecture to support subsequent scale-up. We scale RL compute along three dimensions: (1) larger batches and higher throughput, with an asynchronous training that consumes 1,568 samples and 2.7-3.7B tokens per step at context lengths of up to 1M; (2) more diverse and complex environments, spanning code, general, visual, and cyber domains under a mixture of agent harnesses; and (3) more grader compute, via groupwise agentic grading that yields more accurate reward signals for long-horizon tasks and steers the model towards shorter, more token-efficient solutions. To keep training stable at scale, we freeze the MoE router and establish a multi-layer defense against reward hacking. We further build infrastructure for mixed-task agentic RL, including a unified trajectory representation, high-concurrency multi-framework rollout, decoupled control and data planes, and training-inference consistency. We open-source the training dynamics, RL environments, and RL framework to facilitate reproduction and further research on scaled RL and model self-improvement.
Read the full abstract

All authors Xiaomi LLM-Core Team: Zongming Qiao, Ziyue Hua, Zirui Ou, Zihao Yue, Zihan Jiang, Zhuo Huang, Zhiyang Chen, Zhixian Zheng, Zhipeng Xu, Zhengrui Ma, Yuyang Hu, Yuhang Dong, Yuechen Zhang, Yudong Wang, Yuanxin Liu, Yixin Yang, Yishuo Cai, Yikai Zhao, Yihan Yan, Yifan Zhang, Yifan Song, Xiyu Wei, Xing Zhang, Xin Zhang, Xiaoqian Liu, Xiaodong Ji, Xiangwei Deng, Xueyu Guo, Wenhan Ma, Weimin Xiong, Weikun Wang, Weiji Zhuang, Shuo Liu, Shuhuai Ren, Shuhao Gu, Shimao Chen, Shijie Cao, Shihua Yu, Shicheng Li, Shengjie Zhou, Shaolei Zhang, Rang Li, Qiying Wang, Qingkai Fang, Qianli Chen, Minzheng Wang, Liwen Wang, Linli Yao, Linghao Zhang, Liangyu Cheng, Liang Zhao, Lei Li, Jinhao Dong, Jinyu Xiang, Jianyu Wei, Jiangshan Duo, Huaqiu Liu, Huanjie Fan, Hongyi Guan, Hongshen Xu, Hao Tian, Hanyu Li, Hailin Zhang, Gang Wang, Fuli Luo, Feng Wei, Dong Zhang, Dawei Zhu, Chiheng Lou, Chenhong He, Chenhao He, Chenghua Liu, Bowen Ye, Bowen Shen, Boshen Xu, Bo Yang, Bingquan Xia, Bangjun Xiao, Baixuan Xu, Zhouxiang Mao, Zhiyang Zhang, Zhixiang Xu, Zhenru Lin, Zhengju Tang, Zhaojun Huang, Yuzhe Weng, Yuxing Xiang, Yuxiao Li, Yuheng Yang, Yuhang Wang, Yuchen Liu, Yuanyuan Tian, Yuanliang Dong, Yu Cheng, Yongzhe He, Yongshun Liang, Yong Wang, Yiyan Wang, Yitian Gong, Yijie Zhang, Yanshu Xin, Xun Zhang, Xingjian Zhao, Wenyu Yang, Wenshan Huang, Wenhao Li, Tingwei Huang, Tianyu Yu, Tianyang Lu, Taoyu Yang, Sinan Du, Shutong Tian, Shulin Du, Shengfan Wang, Shanchuan Fang, Qihao Zhang, Qibin Yang, Qian Yu, Qian Tu, Pengrong Xie, Peipei Wang, Peidian Li, Minkun Guo, Mingchen Shao, Luohan Gao, Lijie Wang, Liang Shi, Kaiqi Chen, Kaiming Liu, Kaifei Wang, Kai Yang, Jinlong Xue, Jiechen Zhang, Jiaxuan Liu, Hongxu An, Hao Peng, Hanglong Lü, Guonan Wang, Feiyu Yang, Fanyu Cao, Fangyue Liu, Fan Cui, Cong Wang, Chun Chen, Chenxu Bai, Chengxuan Zhu, Chenghua Wang, Boyi Zeng

Computation and Language

When History Helps and Hurts: Selective History Use across Multimodal Turns

Shuoyang Sun, Kerui Gu, Hao Fang et al. (+2 more)

Reliable multimodal interaction depends on selective use of conversational history: an earlier question may remain relevant while its previous answer is outdated, whereas a current request may depend on historical evidence despite conflicting new observations. Existing multi-turn evaluations rarely separate these history-use demands from underlying question difficulty. To address this gap, we introduce ReTurn, a benchmark of 7,000 base tasks spanning visual and audio evidence for evaluating selective history use. For task-carrying history, Reconfirm/Reground require applying a historical question to current media while varying historical agreement; for evidence-carrying history, Retrieve/Rebind require answering a current question using historical media while varying current-media competition. Each pair preserves the target question, media, and answer. Tasks support open-ended and multiple-choice evaluation, with matched single-turn counterparts serving as answerability references. Across 13 omni-modal, vision-language, and audio-language models, median model-level open-ended accuracy falls from 93.7% with direct input to 72.3% in conversation. Behavioral probes show that high question recall can coexist with weaker task application, while competing media can redirect answers away from historical targets. Supervised adaptation yields only partial gains. ReTurn provides a controlled framework for assessing whether multimodal models select and use the historical information required by each request.
Read the full abstract

All authors Shuoyang Sun, Kerui Gu, Hao Fang, Shaoli Huang, Bin Chen

Computation and Language · Cross-listed from cs.IR

Project Greenhouse: Progress Toward Fully Open and Sovereign Agentic Search

Jimmy Lin, Sahel Sharifymoghaddam, Lingwei Gu et al. (+1 more)

Project Greenhouse represents our exploration of a simple thesis: We believe that it is possible to build fully open and sovereign models for agentic search with only modest computational resources. As a first milestone, we describe how to build a competitive pointwise decoder-only reranker using a simple two-step recipe comprising pre-training from scratch followed by supervised fine-tuning, starting only from commonly available datasets. Contrary to the dominant approach in the literature, we do not rely on existing open-weight backbones from third parties, and thus we are fully in control of model training, from end to end. We were able to accomplish the bulk of our experiments using no more than a handful of GPUs. This report articulates the importance and benefits of our approach, and we share artifacts that enable transparent, independent reproduction of all aspects of model training. Beyond data, code, and configurations that capture our efforts, we also release checkpoints for our family of Gaggle models, demonstrating the feasibility of our approach and providing a first step toward validating our broader thesis.
Read the full abstract

All authors Jimmy Lin, Sahel Sharifymoghaddam, Lingwei Gu, Nour Jedidi

Computation and Language

Event-Centric Memory with Query-Aware Graph Augmentation for Long-Term Conversational Agents

Yichen Liu, Chunfeng Yuan, Haowei Liu et al. (+5 more)

For persistent and personalized conversational agents, memory systems can enable them to remember, update, and reason over long histories by storing past interactions and retrieving relevant information. Existing memory systems typically follow two paradigms: flat-structured memory and graph-based memory. The former is lightweight but leaves event relations and state updates implicit, while the latter explicitly models memory structure but incurs additional construction cost and introduces irrelevant relations over long histories. To address these limitations, we propose QGMem, a novel memory construction and activation framework motivated by human memory, in which experience is organized into events and query-relevant events are modeled by graph as working memory. QGMem converts long dialogue histories into event-indexed atomic memory units that preserve individual experiences and consolidates related units into dynamic memory traces that retain state trajectories and current states. When a query arrives, hybrid memory retrieval gathers complementary candidate memories, and query-aware reranking activates the most relevant units as a compact working memory. To expose relational dependencies in the working memory and support conflict-aware reasoning, QGMem organizes the working memory as a local graph, which is then encoded as a graph token and provided to the LLM together with the textual working memory to improve evidence utilization during answer generation. Experiments across six benchmarks validate the framework and show consistent gains in retrieval, multi-hop evidence composition, conflict resolution, and ultra-long dialogue reasoning with compact contexts and moderate inference cost.
Read the full abstract

All authors Yichen Liu, Chunfeng Yuan, Haowei Liu, Wenjuan Li, Zefeng Lin, Bing Li, Xu Chen, Weiming Hu

Computation and Language

Not Every Change Is Necessary: Recoverable Drift in Large Language Model Unlearning

Xunlei Chen, Qinghui Gong, Jingkun Xue et al. (+3 more)

Machine unlearning in large language models aims to remove unwanted knowledge while preserving the model's remaining capabilities. Although existing methods use retention objectives or restrict where edits occur, achieving the desired forgetting level can still leave collateral changes that impair non-target behavior. Our recovery comparisons suggest that some of these changes can be reversed while preserving observed forgetting performance. In this work, we present Propose-Then-Project Unlearning (PTP-U), a framework that combines targeted forgetting with the recovery of non-target capabilities. PTP-U first applies local analytic edits to weaken target knowledge associations, then aligns non-target output distributions with those of the original model to recover capabilities while maintaining fixed forgetting constraints. Both stages serve a common goal: satisfying the forgetting requirements while preserving fluent generation and performance on non-target tasks. Across three benchmarks, PTP-U achieves the strongest forgetting-retention trade-off among evaluated methods, reaching 81.22%-91.03% forgetting while preserving 94.20% non-target utility on average. At matched forgetting, PTP-U consistently retains higher non-target utility.
Read the full abstract

All authors Xunlei Chen, Qinghui Gong, Jingkun Xue, Qihe Liu, Shijie Zhou, Fei Ye

Computation and Language

Can Decision Models Understand Stance? Evaluating Jev Against General-Purpose LLMs

Xing Li, Jinzhong Ning, Yijia Zhang et al. (+2 more)

Stance detection requires identifying an author's attitude toward a given target, sometimes based on conversational context. Jev, a specialized decision model designed for structured decision-making, offers an alternative to general-purpose large language models (LLMs). In this work, we evaluate Jev on two stance detection datasets, VAST (English texts) and ZS-CSD (Chinese conversations), comparing it with four general-purpose LLMs and two fine-tuned models. Results show that Jev achieves competitive performance on VAST, matching GPT-5.6 and outperforming the other general-purpose LLMs. However, it falls behind stronger LLMs on ZS-CSD, particularly in distinguishing favor from against. Further analysis suggests that this limitation may be related to understanding reply relationships and stance direction rather than conversation length alone. These findings highlight both the potential and limitations of Jev for stance detection.
Read the full abstract

All authors Xing Li, Jinzhong Ning, Yijia Zhang, Liang Yang, Hongfei Lin

Computation and Language

Forms of LLM-Integrated Applications from LLM-Chats to Autonomous AI Agent System

Irene Weber

Large language models (LLMs) are increasingly embedded as components in software systems, marketed under labels such as chatbot, copilot, retrieval-augmented generation, workflow, coding agent and AI agent. Whether these labels denote genuine architectural forms or serve as branding has not been assessed systematically. In the sources surveyed, labels do carry architectural content, most clearly in vendor usage: copilot denotes a router-worker architecture operating a host application under step-by-step user confirmation, while the more recent shift to the label agent coincides with AI-planned multi-step execution of which the user sees only the outcome. The coding agents of four major providers share one architecture, a reason-and-act loop delegating to subagents. This survey describes seven recurring forms—LLM chats, custom agents, retrieval-augmented generation (RAG), AI-enhanced workflows, copilots, coding agents, and, in part, agentic RAG—in a common vocabulary of agents and tools. Each is characterized along four structural dimensions (agentic RAG only partially): the architectural pattern, the control of execution and the point of user intervention, the number of agent calls per task, and tool use. An illustrative corpus of 22 systems from research publications and vendor documentation grounds the descriptions and shows where they reach their limit.
Read the full abstract

Computation and Language · Cross-listed from cs.LG

GRPODropout: Less is More for Online Reinforcement Learning Rollouts

Hexuan Deng, Zihao Yan, Xuebo Liu et al. (+8 more)

Reinforcement learning (RL) methods such as GRPO substantially improve large language model reasoning but often suffer from policy entropy collapse: the loss of sampling diversity weakens exploration and limits further improvement. Existing methods address this issue either through algorithm-level interventions, such as reward modification and entropy/KL regularization, or through token-level reweighting. We investigate a complementary perspective: entropy collapse can also be mitigated by changing which generated rollouts contribute to policy updates. Under the same sampling budget, not all rollouts contribute positively to an update, and selectively excluding some can improve learning. To address this, we propose GRPODropout: before the standard update, we use a simple strategy that selectively removes a small number of high-probability positive-advantage rollouts and recenters the retained advantages. To motivate this design, we develop a rollout-level theoretical analysis that guides method design and threshold selection. The method changes only rollout usage, and adds negligible computational overhead. Experiments show higher accuracy than original GRPO and higher actor entropy while using fewer rollout samples for updates, illustrating "less is more." This work provides insight into RL rollout usage: removing some rollouts can improve performance. Code is available at https://github.com/hexuandeng/GRPODropout/.
Read the full abstract

All authors Hexuan Deng, Zihao Yan, Xuebo Liu, Shuo Nie, Yue Wang, Chen Wang, Zhaohua Zhang, Tianwen Jiang, Qiuyong Xiao, Jihong Zhang, Min Zhang