Showing 1–50 of 357 results for author: Du, L

TAMISeg: Text-Aligned Multi-scale Medical Image Segmentation with Semantic Encoder Distillation

Authors: Qiang Gao, Yi Wang, Yong Zhang, Yong Li, Yongbing Deng, Lan Du, Cunjian Chen

Abstract: Medical image segmentation remains challenging due to limited fine-grained annotations, complex anatomical structures, and image degradation from noise, low contrast, or illumination variation. We propose TAMISeg, a text-guided segmentation framework that incorporates clinical language prompts and semantic distillation as auxiliary semantic cues to enhance visual understanding and reduce reliance… ▽ More Medical image segmentation remains challenging due to limited fine-grained annotations, complex anatomical structures, and image degradation from noise, low contrast, or illumination variation. We propose TAMISeg, a text-guided segmentation framework that incorporates clinical language prompts and semantic distillation as auxiliary semantic cues to enhance visual understanding and reduce reliance on pixel-level fine-grained annotations. TAMISeg integrates three core components: a consistency-aware encoder pretrained with strong perturbations for robust feature extraction, a semantic encoder distillation module with supervision from a frozen DINOv3 teacher to enhance semantic discriminability, and a scale-adaptive decoder that segments anatomical structures across different spatial scales. Experiments on the Kvasir-SEG, MosMedData+, and QaTa-COV19 datasets demonstrate that TAMISeg consistently outperforms existing uni-modal and multi-modal methods in both qualitative and quantitative evaluations. Code will be made publicly available at https://github.com/qczggaoqiang/TAMISeg. △ Less

Submitted 12 April, 2026; originally announced April 2026.

Comments: Accepted by IEEE International Conference on Multimedia and Expo (ICME), 2026

NEURA: A Unified and Retargetable Compilation Framework for Coarse-Grained Reconfigurable Architectures

Authors: Shangkun Li, Jinming Ge, Diyuan Tao, Zeyu Li, Jiawei Liang, Linfeng Du, Jiang Xu, Wei Zhang, Cheng Tan

Abstract: Coarse-Grained Reconfigurable Architectures (CGRAs) are a promising and versatile accelerator platform, offering a balance between the performance and efficiency of specialized accelerators and the software programmability. However, their full potential is severely hindered by control flow in accelerated kernels, as the control flow (e.g., loops, branches) is fundamentally incompatible with the pa… ▽ More Coarse-Grained Reconfigurable Architectures (CGRAs) are a promising and versatile accelerator platform, offering a balance between the performance and efficiency of specialized accelerators and the software programmability. However, their full potential is severely hindered by control flow in accelerated kernels, as the control flow (e.g., loops, branches) is fundamentally incompatible with the parallel, data-driven CGRA fabric. Prior strategies to resolve this mismatch in CGRA kernel acceleration are either inefficient, sacrificing performance for generality, or lack generality due to the difficulty of adapting them across different execution models. Thus, a general and unified solution for efficient CGRA kernel acceleration remains elusive. This paper introduces NEURA, a unified and retargetable compilation framework that systematically resolves the control-dataflow mismatch in CGRAs. NEURA's core innovation is a novel, pure dataflow intermediate representation (IR) built on a predicated type system. In this IR, control contexts are embedded as a predicate within each data, making control an intrinsic property of data. This mechanism enables NEURA to systematically flatten complex control flow into a single unified dataflow graph. This unified representation decouples kernel representation from hardware, empowering NEURA to retarget diverse CGRAs with different execution models and microarchitectural features. When targeted to a high-performance spatio-temporal CGRA, NEURA delivers a 2.20x speedup on kernel benchmarks and up to 2.71x geometric mean speedup on real-world applications over state-of-the-art (SOTA) high-performance baselines. It also provides a competitive solution against the SOTA low-power CGRA when retargeted to a spatial-only CGRA. NEURA is open-source and available at https://github.com/coredac/neura. △ Less

Submitted 5 April, 2026; originally announced April 2026.

Comments: Accepted by PLDI 2026

Beyond Message Passing: A Semantic View of Agent Communication Protocols

Authors: Dun Yuan, Fuyuan Lyu, Ye Yuan, Weixu Zhang, Bowei He, Jiayi Geng, Linfeng Du, Zipeng Sun, Yankai Chen, Changjiang Han, Jikun Kang, Xi Chen, Haolun Wu, Xue Liu

Abstract: Agent communication protocols are becoming critical infrastructure for large language model (LLM) systems that must use tools, coordinate with other agents, and operate across heterogeneous environments. This work presents a human-inspired perspective on this emerging landscape by organizing agent communication into three layers: communication, syntactic, and semantic. Under this framework, we sys… ▽ More Agent communication protocols are becoming critical infrastructure for large language model (LLM) systems that must use tools, coordinate with other agents, and operate across heterogeneous environments. This work presents a human-inspired perspective on this emerging landscape by organizing agent communication into three layers: communication, syntactic, and semantic. Under this framework, we systematically analyze 18 representative protocols and compare how they support reliable transport, structured interaction, and meaning-level coordination. Our analysis shows a clear imbalance in current protocol design. Most protocols provide increasingly mature support for transport, streaming, schema definition, and lifecycle management, but offer limited protocol-level mechanisms for clarification, context alignment, and verification. As a result, semantic responsibilities are often pushed into prompts, wrappers, or application-specific orchestration logic, creating hidden interoperability and maintenance costs. To make this gap actionable, we further identify major forms of technical debt in today's protocol ecosystem and distill practical guidance for selecting protocols under different deployment settings. We conclude by outlining a research agenda for interoperable, secure, and semantically robust agent ecosystems that move beyond message passing toward shared understanding. △ Less

Submitted 13 April, 2026; v1 submitted 29 March, 2026; originally announced April 2026.

Key-Embedded Privacy for Decentralized AI in Biomedical Omics

Authors: Rongyu Zhang, Hongyu Dong, Gaole Dai, Ziqi Qiao, Shenli Zheng, Yuan Zhang, Aosong Cheng, Xiaowei Chi, Jincai Luo, Pin Li, Li Du, Dan Wang, Yuan Du, Xudong Xing, Jianxu Chen, Shanghang Zhang

Abstract: The rapid adoption of data-driven methods in biomedicine has intensified concerns over privacy, governance, and regulation, limiting raw data sharing and hindering the assembly of representative cohorts for clinically relevant AI. This landscape necessitates practical, efficient privacy solutions, as cryptographic defenses often impose heavy overhead and differential privacy can degrade performanc… ▽ More The rapid adoption of data-driven methods in biomedicine has intensified concerns over privacy, governance, and regulation, limiting raw data sharing and hindering the assembly of representative cohorts for clinically relevant AI. This landscape necessitates practical, efficient privacy solutions, as cryptographic defenses often impose heavy overhead and differential privacy can degrade performance, leading to sub-optimal outcomes in real-world settings. Here, we present a lightweight federated learning method, INFL, based on Implicit Neural Representations that addresses these challenges. Our approach integrates plug-and-play, coordinate-conditioned modules into client models, embeds a secret key directly into the architecture, and supports seamless aggregation across heterogeneous sites. Across diverse biomedical omics tasks, including cohort-scale classification in bulk proteomics, regression for perturbation prediction in single-cell transcriptomics, and clustering in spatial transcriptomics and multi-omics with both public and private data, we demonstrate that INFL achieves strong, controllable privacy while maintaining utility, preserving the performance necessary for downstream scientific and clinical applications. △ Less

Submitted 30 March, 2026; originally announced March 2026.

MR-ImagenTime: Multi-Resolution Time Series Generation through Dual Image Representations

Authors: Xianyong Xu, Yuanjun Zuo, Zhihong Huang, Yihan Qin, Haoxian Xu, Leilei Du, Haotian Wang

Abstract: Time series forecasting is vital across many domains, yet existing models struggle with fixed-length inputs and inadequate multi-scale modeling. We propose MR-CDM, a framework combining hierarchical multi-resolution trend decomposition, an adaptive embedding mechanism for variable-length inputs, and a multi-scale conditional diffusion process. Evaluations on four real-world datasets demonstrate th… ▽ More Time series forecasting is vital across many domains, yet existing models struggle with fixed-length inputs and inadequate multi-scale modeling. We propose MR-CDM, a framework combining hierarchical multi-resolution trend decomposition, an adaptive embedding mechanism for variable-length inputs, and a multi-scale conditional diffusion process. Evaluations on four real-world datasets demonstrate that MR-CDM significantly outperforms state-of-the-art baselines (e.g., CSDI, Informer), reducing MAE and RMSE by approximately 6-10 to a certain degree. △ Less

Submitted 7 April, 2026; v1 submitted 30 March, 2026; originally announced March 2026.

A Switch-Centric In-Network Architecture for Accelerating LLM Inference in Shared-Memory Network

Authors: Aojie Jiang, Kang Zhu, Zhiheng Zhang, Zhengxu Su, Juntao Liu, Yuan Du, Li Du

Abstract: In-network computing techniques, exemplified by NVLink SHARP (NVLS), offer a promising approach to addressing the communication bottlenecks in LLM inference by offloading collective operations such as All-Reduce to switches. However, the accelerator-centric architecture of NVLS suffers from two fundamental limitations: 1) it relies on GPU load instructions to trigger in-switch reduction, which mea… ▽ More In-network computing techniques, exemplified by NVLink SHARP (NVLS), offer a promising approach to addressing the communication bottlenecks in LLM inference by offloading collective operations such as All-Reduce to switches. However, the accelerator-centric architecture of NVLS suffers from two fundamental limitations: 1) it relies on GPU load instructions to trigger in-switch reduction, which means that the data reduced in the switch must be transferred back to the initiating GPU rather than being broadcast directly, thereby introducing unnecessary communication overhead; 2) due to its architectural constraints, NVLS cannot offload operators that are not decomposable into memory-semantic instructions, such as the in-network quantization (INQ) proposed in this work. As a result, All-Reduce in NVLS during inference still operates at 16-bit precision, leading to substantial bandwidth waste. To address these limitations, we propose SCIN, the first switch-centric in-network architecture for multi-accelerator shared-memory networks, enabling both low-latency and high-bandwidth All-Reduce. Specifically, we introduce an in-switch accelerator (ISA) capable of directly accessing the memory regions in attached accelerators for in-network processing, together with a co-designed communication fabric that enables such access with negligible protocol overhead. SCIN delivers lower All-Reduce latency than NVLS by eliminating redundant data movement. Moreover, SCIN enables INQ for All-Reduce, reducing its precision to 8 bits and nearly doubling bandwidth with negligible accuracy loss. We also present a multi-FPGA prototype of SCIN to validate its feasibility and effectiveness. Simulation results for an 8-GPU system show that our design accelerates All-Reduce by up to 8.7x for small messages and 3.8x for large messages, yielding up to 1.74x TTFT speedup and 1.34x TPOT speedup on LLaMA-2 models. △ Less

Submitted 8 April, 2026; v1 submitted 30 March, 2026; originally announced March 2026.

Toward Physically Consistent Driving Video World Models under Challenging Trajectories

Authors: Jiawei Zhou, Zhenxin Zhu, Lingyi Du, Linye Lyu, Lijun Zhou, Zhanqian Wu, Hongcheng Luo, Zhuotao Tian, Bing Wang, Guang Chen, Hangjun Ye, Haiyang Sun, Yu Li

Abstract: Video generation models have shown strong potential as world models for autonomous driving simulation. However, existing approaches are primarily trained on real-world driving datasets, which mostly contain natural and safe driving scenarios. As a result, current models often fail when conditioned on challenging or counterfactual trajectories-such as imperfect trajectories generated by simulators… ▽ More Video generation models have shown strong potential as world models for autonomous driving simulation. However, existing approaches are primarily trained on real-world driving datasets, which mostly contain natural and safe driving scenarios. As a result, current models often fail when conditioned on challenging or counterfactual trajectories-such as imperfect trajectories generated by simulators or planning systems-producing videos with severe physical inconsistencies and artifacts. To address this limitation, we propose PhyGenesis, a world model designed to generate driving videos with high visual fidelity and strong physical consistency. Our framework consists of two key components: (1) a physical condition generator that transforms potentially invalid trajectory inputs into physically plausible conditions, and (2) a physics-enhanced video generator that produces high-fidelity multi-view driving videos under these conditions. To effectively train these components, we construct a large-scale, physics-rich heterogeneous dataset. Specifically, in addition to real-world driving videos, we generate diverse challenging driving scenarios using the CARLA simulator, from which we derive supervision signals that guide the model to learn physically grounded dynamics under extreme conditions. This challenging-trajectory learning strategy enables trajectory correction and promotes physically consistent video generation. Extensive experiments demonstrate that PhyGenesis consistently outperforms state-of-the-art methods, especially on challenging trajectories. Our project page is available at: https://wm-research.github.io/PhyGenesis/. △ Less

Submitted 1 April, 2026; v1 submitted 25 March, 2026; originally announced March 2026.

LARFT: Closing the Cognition-Action Gap for Length Instruction Following in Large Language Models

Authors: Wei Zhang, Lintong Du, Yuanhe Zhang, Zhenhong Zhou, Kun Wang, Li Sun, Sen Su

Abstract: Despite the strong performance of Large Language Models (LLMs) on complex instruction-following tasks, precise control of output length remains a persistent challenge. Existing methods primarily attempt to enforce length constraints by externally imposing length signals or optimization objectives, while largely overlooking the underlying limitation: the model's intrinsic deficit in length cognitio… ▽ More Despite the strong performance of Large Language Models (LLMs) on complex instruction-following tasks, precise control of output length remains a persistent challenge. Existing methods primarily attempt to enforce length constraints by externally imposing length signals or optimization objectives, while largely overlooking the underlying limitation: the model's intrinsic deficit in length cognition. To address this, we propose LARFT (Length-Aware Reinforcement Fine-Tuning), a training framework that aligns the model's length cognition with its action. Specifically, LARFT integrates length-oriented reinforcement learning with a hindsight length awareness. By transforming on-policy data into hindsight self-awareness tasks where the model learns to identify the actual length of its own generation, LARFT jointly optimizes the model's internal representation of length information and refines its policy to satisfy length constraints, thereby achieving precise and reliable length instruction following. Extensive experiments across four base models demonstrate that LARFT outperforms existing baselines, achieving an average improvement of +20.92 points across three length instruction following benchmarks with only a marginal decline of -1.45 points on four general capability benchmarks. △ Less

Submitted 25 February, 2026; originally announced March 2026.

Comments: 19 pages, 6 figures

Gym-V: A Unified Vision Environment System for Agentic Vision Research

Authors: Fanqing Meng, Lingxiao Du, Jiawei Gu, Jiaqi Liao, Linjie Li, Zijian Wu, Xiangyan Liu, Ziqi Zhao, Mengkang Hu, Zichen Liu, Jiaheng Zhang, Michael Qizhe Shieh

Abstract: As agentic systems increasingly rely on reinforcement learning from verifiable rewards, standardized ``gym'' infrastructure has become essential for rapid iteration, reproducibility, and fair comparison. Vision agents lack such infrastructure, limiting systematic study of what drives their learning and where current models fall short. We introduce \textbf{Gym-V}, a unified platform of 179 procedur… ▽ More As agentic systems increasingly rely on reinforcement learning from verifiable rewards, standardized ``gym'' infrastructure has become essential for rapid iteration, reproducibility, and fair comparison. Vision agents lack such infrastructure, limiting systematic study of what drives their learning and where current models fall short. We introduce \textbf{Gym-V}, a unified platform of 179 procedurally generated visual environments across 10 domains with controllable difficulty, enabling controlled experiments that were previously infeasible across fragmented toolkits. Using it, we find that observation scaffolding is more decisive for training success than the choice of RL algorithm, with captions and game rules determining whether learning succeeds at all. Cross-domain transfer experiments further show that training on diverse task categories generalizes broadly while narrow training can cause negative transfer, with multi-turn interaction amplifying all of these effects. Gym-V is released as a convenient foundation for training environments and evaluation toolkits, aiming to accelerate future research on agentic VLMs. △ Less

Submitted 8 April, 2026; v1 submitted 16 March, 2026; originally announced March 2026.

AutoSkill: Experience-Driven Lifelong Learning via Skill Self-Evolution

Authors: Yutao Yang, Junsong Li, Qianjun Pan, Bihao Zhan, Yuxuan Cai, Lin Du, Jie Zhou, Kai Chen, Qin Chen, Xin Li, Bo Zhang, Liang He

Abstract: In practical LLM applications, users repeatedly express stable preferences and requirements, such as reducing hallucinations, following institutional writing conventions, or avoiding overly technical wording, yet such interaction experience is seldom consolidated into reusable knowledge. Consequently, LLM agents often fail to accumulate personalized capabilities across sessions. We present AutoSki… ▽ More In practical LLM applications, users repeatedly express stable preferences and requirements, such as reducing hallucinations, following institutional writing conventions, or avoiding overly technical wording, yet such interaction experience is seldom consolidated into reusable knowledge. Consequently, LLM agents often fail to accumulate personalized capabilities across sessions. We present AutoSkill, an experience-driven lifelong learning framework that enables LLM agents to automatically derive, maintain, and reuse skills from dialogue and interaction traces. AutoSkill abstracts skills from user experience, supports their continual self-evolution, and dynamically injects relevant skills into future requests without retraining the underlying model. Designed as a model-agnostic plugin layer, it is compatible with existing LLMs and introduces a standardized skill representation for sharing and transfer across agents, users, and tasks. In this way, AutoSkill turns ephemeral interaction experience into explicit, reusable, and composable capabilities. This paper describes the motivation, architecture, skill lifecycle, and implementation of AutoSkill, and positions it with respect to prior work on memory, retrieval, personalization, and agentic systems. AutoSkill highlights a practical and scalable path toward lifelong personalized agents and personal digital surrogates. △ Less

Submitted 4 March, 2026; v1 submitted 1 March, 2026; originally announced March 2026.

FPPS: An FPGA-Based Point Cloud Processing System

Authors: Xiaofeng Zhou, Linfeng Du, Hanwei Fan, Wei Zhang

Abstract: Point cloud processing is a computational bottleneck in autonomous driving systems, especially for real-time applications, while energy efficiency remains a critical system constraint. This work presents FPPS, an FPGA-accelerated point cloud processing system designed to optimize the iterative closest point (ICP) algorithm, a classic cornerstone of 3D localization and perception pipelines. Evaluat… ▽ More Point cloud processing is a computational bottleneck in autonomous driving systems, especially for real-time applications, while energy efficiency remains a critical system constraint. This work presents FPPS, an FPGA-accelerated point cloud processing system designed to optimize the iterative closest point (ICP) algorithm, a classic cornerstone of 3D localization and perception pipelines. Evaluated on the widely used KITTI benchmark dataset, the proposed system achieves up to 35$\times$ (and a runtime-weighted average of 15.95x) speedup over a state-of-the-art CPU baseline while maintaining equivalent registration accuracy. Notably, the design improves average power efficiency by 8.58x, offering a compelling balance between performance and energy consumption. These results position FPPS as a viable solution for resource-constrained embedded autonomous platforms where both latency and power are key design priorities. △ Less

Submitted 27 February, 2026; originally announced February 2026.

Extracting Consumer Insight from Text: A Large Language Model Approach to Emotion and Evaluation Measurement

Authors: Stephan Ludwig, Peter J. Danaher, Xiaohao Yang, Yu-Ting Lin, Ehsan Abedin, Dhruv Grewal, Lan Du

Abstract: Accurately measuring consumer emotions and evaluations from unstructured text remains a core challenge for marketing research and practice. This study introduces the Linguistic eXtractor (LX), a fine-tuned, large language model trained on consumer-authored text that also has been labeled with consumers' self-reported ratings of 16 consumption-related emotions and four evaluation constructs: trust,… ▽ More Accurately measuring consumer emotions and evaluations from unstructured text remains a core challenge for marketing research and practice. This study introduces the Linguistic eXtractor (LX), a fine-tuned, large language model trained on consumer-authored text that also has been labeled with consumers' self-reported ratings of 16 consumption-related emotions and four evaluation constructs: trust, commitment, recommendation, and sentiment. LX consistently outperforms leading models, including GPT-4 Turbo, RoBERTa, and DeepSeek, achieving 81% macro-F1 accuracy on open-ended survey responses and greater than 95% accuracy on third-party-annotated Amazon and Yelp reviews. An application of LX to online retail data, using seemingly unrelated regression, affirms that review-expressed emotions predict product ratings, which in turn predict purchase behavior. Most emotional effects are mediated by product ratings, though some emotions, such as discontent and peacefulness, influence purchase directly, indicating that emotional tone provides meaningful signals beyond star ratings. To support its use, a no-code, cost-free, LX web application is available, enabling scalable analyses of consumer-authored text. In establishing a new methodological foundation for consumer perception measurement, this research demonstrates new methods for leveraging large language models to advance marketing research and practice, thereby achieving validated detection of marketing constructs from consumer data. △ Less

Submitted 16 February, 2026; originally announced February 2026.

DWBench: Holistic Evaluation of Watermark for Dataset Copyright Auditing

Authors: Xiao Ren, Xinyi Yu, Linkang Du, Min Chen, Yuanchao Shu, Zhou Su, Yunjun Gao, Zhikun Zhang

Abstract: The surging demand for large-scale datasets in deep learning has heightened the need for effective copyright protection, given the risks of unauthorized use to data owners. Although the dataset watermark technique holds promise for auditing and verifying usage, existing methods are hindered by inconsistent evaluations, which impede fair comparisons and assessments of real-world viability. To addre… ▽ More The surging demand for large-scale datasets in deep learning has heightened the need for effective copyright protection, given the risks of unauthorized use to data owners. Although the dataset watermark technique holds promise for auditing and verifying usage, existing methods are hindered by inconsistent evaluations, which impede fair comparisons and assessments of real-world viability. To address this gap, we propose a two-layer taxonomy that categorizes methods by implementation (model-based vs. model-free injection; model-behavior vs. model-message verification), offering a structured framework for cross-task analysis. Then, we develop DWBench, a unified benchmark and open-source toolkit for systematically evaluating image dataset watermark techniques in classification and generation tasks. Using DWBench, we assess 25 representative methods under standardized conditions, perturbation-based robustness tests, multi-watermark coexistence, and multi-user interference. In addition to reporting the results of four commonly used metrics, we present the results of two new metrics: sample significance for fine-grained watermark distinguishability and verification success rate for dataset-level auditing, which enable accurate and reproducible benchmarking. Key findings reveal inherent trade-offs: no single method dominates all scenarios; classification and generation tasks require specialized approaches; and existing techniques exhibit instability at low watermark rates and in realistic multi-user settings, with elevated false positives or performance declines. We hope that DWBench can facilitate advances in watermark reliability and practicality, thus strengthening copyright safeguards in the face of widespread AI-driven data exploitation. △ Less

Submitted 13 February, 2026; originally announced February 2026.

Comments: 19 pages, 4 figures

LLaDA2.1: Speeding Up Text Diffusion via Token Editing

Authors: Tiwei Bie, Maosong Cao, Xiang Cao, Bingsen Chen, Fuyuan Chen, Kun Chen, Lun Du, Daozhuo Feng, Haibo Feng, Mingliang Gong, Zhuocheng Gong, Yanmei Gu, Jian Guan, Kaiyuan Guan, Hongliang He, Zenan Huang, Juyong Jiang, Zhonghui Jiang, Zhenzhong Lan, Chengxi Li, Jianguo Li, Zehuan Li, Huabin Liu, Lin Liu, Guoshan Lu , et al. (25 additional authors not shown)

Abstract: While LLaDA2.0 showcased the scaling potential of 100B-level block-diffusion models and their inherent parallelization, the delicate equilibrium between decoding speed and generation quality has remained an elusive frontier. Today, we unveil LLaDA2.1, a paradigm shift designed to transcend this trade-off. By seamlessly weaving Token-to-Token (T2T) editing into the conventional Mask-to-Token (M2T)… ▽ More While LLaDA2.0 showcased the scaling potential of 100B-level block-diffusion models and their inherent parallelization, the delicate equilibrium between decoding speed and generation quality has remained an elusive frontier. Today, we unveil LLaDA2.1, a paradigm shift designed to transcend this trade-off. By seamlessly weaving Token-to-Token (T2T) editing into the conventional Mask-to-Token (M2T) scheme, we introduce a joint, configurable threshold-decoding scheme. This structural innovation gives rise to two distinct personas: the Speedy Mode (S Mode), which audaciously lowers the M2T threshold to bypass traditional constraints while relying on T2T to refine the output; and the Quality Mode (Q Mode), which leans into conservative thresholds to secure superior benchmark performances with manageable efficiency degrade. Furthering this evolution, underpinned by an expansive context window, we implement the first large-scale Reinforcement Learning (RL) framework specifically tailored for dLLMs, anchored by specialized techniques for stable gradient estimation. This alignment not only sharpens reasoning precision but also elevates instruction-following fidelity, bridging the chasm between diffusion dynamics and complex human intent. We culminate this work by releasing LLaDA2.1-Mini (16B) and LLaDA2.1-Flash (100B). Across 33 rigorous benchmarks, LLaDA2.1 delivers strong task performance and lightning-fast decoding speed. Despite its 100B volume, on coding tasks it attains an astounding 892 TPS on HumanEval+, 801 TPS on BigCodeBench, and 663 TPS on LiveCodeBench. △ Less

Submitted 13 February, 2026; v1 submitted 9 February, 2026; originally announced February 2026.

Comments: 11 pages, 3 figures

LLMs Know More About Numbers than They Can Say

Authors: Fengting Yuchi, Li Du, Jason Eisner

Abstract: Although state-of-the-art LLMs can solve math problems, we find that they make errors on numerical comparisons with mixed notation: "Which is larger, $5.7 \times 10^2$ or $580$?" This raises a fundamental question: Do LLMs even know how big these numbers are? We probe the hidden states of several smaller open-source LLMs. A single linear projection of an appropriate hidden layer encodes the log-ma… ▽ More Although state-of-the-art LLMs can solve math problems, we find that they make errors on numerical comparisons with mixed notation: "Which is larger, $5.7 \times 10^2$ or $580$?" This raises a fundamental question: Do LLMs even know how big these numbers are? We probe the hidden states of several smaller open-source LLMs. A single linear projection of an appropriate hidden layer encodes the log-magnitudes of both kinds of numerals, allowing us to recover the numbers with relative error of about 2.3% (on restricted synthetic text) or 19.06% (on scientific papers). Furthermore, the hidden state after reading a pair of numerals encodes their ranking, with a linear classifier achieving over 90% accuracy. Yet surprisingly, when explicitly asked to rank the same pairs of numerals, these LLMs achieve only 50-70% accuracy, with worse performance for models whose probes are less effective. Finally, we show that incorporating the classifier probe's log-loss as an auxiliary objective during finetuning brings an additional 3.22% improvement in verbalized accuracy over base models, demonstrating that improving models' internal magnitude representations can enhance their numerical reasoning capabilities. Our code is available at https://github.com/VCY019/Numeracy-Probing. △ Less

Submitted 17 February, 2026; v1 submitted 7 February, 2026; originally announced February 2026.

Comments: EACL 2026 (Oral), camera-ready version with GitHub link

(Computer) Vision in Action: Comparing Remote Sighted Assistance and a Multimodal Voice Agent in Inspection Sequences

Authors: Damien Rudaz, Barbara Nino Carreras, Sara Merlino, Brian L. Due, Barry Brown

Abstract: Does human-AI assistance unfold in the same way as human-human assistance? This research explores what can be learned from the expertise of blind individuals and sighted volunteers to inform the design of multimodal voice agents and address the enduring challenge of proactivity. Drawing on granular analysis of two representative fragments from a larger corpus, we contrast the practices co-produced… ▽ More Does human-AI assistance unfold in the same way as human-human assistance? This research explores what can be learned from the expertise of blind individuals and sighted volunteers to inform the design of multimodal voice agents and address the enduring challenge of proactivity. Drawing on granular analysis of two representative fragments from a larger corpus, we contrast the practices co-produced by an experienced human remote sighted assistant and a blind participant-as they collaborate to find a stain on a blanket over the phone-with those achieved when the same participant worked with a multimodal voice agent on the same task, a few moments earlier. This comparison enables us to specify precisely which fundamental proactive practices the agent did not enact in situ. We conclude that, so long as multimodal voice agents cannot produce environmentally occasioned vision-based actions, they will lack a key resource relied upon by human remote sighted assistants. △ Less

Submitted 5 February, 2026; originally announced February 2026.

Comments: Conditionally accepted at CHI 2026, 32 pages, 8 figures

On the Summability Problem of Multivariate Rational Functions in the Mixed Case

Authors: Shaoshi Chen, Lixin Du, Hanqian Fang, Yisen Wang

Abstract: Continuing previous work, this paper focuses on the summability problem of multivariate rational functions in the mixed case in which both shift and $q$-shift operators can appear. Our summability criteria rely on three ingredients including orbital decompositions, Sato's isotropy groups, and difference transformations. This work settles the rational case of the long-term project aimed at developi… ▽ More Continuing previous work, this paper focuses on the summability problem of multivariate rational functions in the mixed case in which both shift and $q$-shift operators can appear. Our summability criteria rely on three ingredients including orbital decompositions, Sato's isotropy groups, and difference transformations. This work settles the rational case of the long-term project aimed at developing algorithms for symbolic summation of multivariate functions. △ Less

Submitted 3 February, 2026; originally announced February 2026.

Kimi K2.5: Visual Agentic Intelligence

Authors: Kimi Team, Tongtong Bai, Yifan Bai, Yiping Bao, S. H. Cai, Yuan Cao, Y. Charles, H. S. Che, Cheng Chen, Guanduo Chen, Huarong Chen, Jia Chen, Jiahao Chen, Jianlong Chen, Jun Chen, Kefan Chen, Liang Chen, Ruijue Chen, Xinhao Chen, Yanru Chen, Yanxu Chen, Yicun Chen, Yimin Chen, Yingjiang Chen, Yuankun Chen , et al. (301 additional authors not shown)

Abstract: We introduce Kimi K2.5, an open-source multimodal agentic model designed to advance general agentic intelligence. K2.5 emphasizes the joint optimization of text and vision so that two modalities enhance each other. This includes a series of techniques such as joint text-vision pre-training, zero-vision SFT, and joint text-vision reinforcement learning. Building on this multimodal foundation, K2.5… ▽ More We introduce Kimi K2.5, an open-source multimodal agentic model designed to advance general agentic intelligence. K2.5 emphasizes the joint optimization of text and vision so that two modalities enhance each other. This includes a series of techniques such as joint text-vision pre-training, zero-vision SFT, and joint text-vision reinforcement learning. Building on this multimodal foundation, K2.5 introduces Agent Swarm, a self-directed parallel agent orchestration framework that dynamically decomposes complex tasks into heterogeneous sub-problems and executes them concurrently. Extensive evaluations show that Kimi K2.5 achieves state-of-the-art results across various domains including coding, vision, reasoning, and agentic tasks. Agent Swarm also reduces latency by up to $4.5\times$ over single-agent baselines. We release the post-trained Kimi K2.5 model checkpoint to facilitate future research and real-world applications of agentic intelligence. △ Less

Submitted 2 February, 2026; originally announced February 2026.

Comments: Kimi K2.5 tech report

MAIN-VLA: Modeling Abstraction of Intention and eNvironment for Vision-Language-Action Models

Authors: Zheyuan Zhou, Liang Du, Zixun Sun, Xiaoyu Zhou, Ruimin Ye, Qihao Chen, Yinda Chen, Lemiao Qiu

Abstract: Despite significant progress in Visual-Language-Action (VLA), in highly complex and dynamic environments that involve real-time unpredictable interactions (such as 3D open worlds and large-scale PvP games), existing approaches remain inefficient at extracting action-critical signals from redundant sensor streams. To tackle this, we introduce MAIN-VLA, a framework that explicitly Models the Abstrac… ▽ More Despite significant progress in Visual-Language-Action (VLA), in highly complex and dynamic environments that involve real-time unpredictable interactions (such as 3D open worlds and large-scale PvP games), existing approaches remain inefficient at extracting action-critical signals from redundant sensor streams. To tackle this, we introduce MAIN-VLA, a framework that explicitly Models the Abstraction of Intention and eNvironment to ground decision-making in deep semantic alignment rather than superficial pattern matching. Specifically, our Intention Abstraction (IA) extracts verbose linguistic instructions and their associated reasoning into compact, explicit semantic primitives, while the Environment Semantics Abstraction (ESA) projects overwhelming visual streams into a structured, topological affordance representation. Furthermore, aligning these two abstract modalities induces an emergent attention-concentration effect, enabling a parameter-free token-pruning strategy that filters out perceptual redundancy without degrading performance. Extensive experiments in open-world Minecraft and large-scale PvP environments (Game for Peace and Valorant) demonstrate that MAIN-VLA sets a new state-of-the-art, which achieves superior decision quality, stronger generalization, and cutting-edge inference efficiency. △ Less

Submitted 2 February, 2026; originally announced February 2026.

What Does Vision Tool-Use Reinforcement Learning Really Learn? Disentangling Tool-Induced and Intrinsic Effects for Crop-and-Zoom

Authors: Yan Ma, Weiyu Zhang, Tianle Li, Linge Du, Xuyang Shen, Pengfei Liu

Abstract: Vision tool-use reinforcement learning (RL) can equip vision-language models with visual operators such as crop-and-zoom and achieves strong performance gains, yet it remains unclear whether these gains are driven by improvements in tool use or evolving intrinsic capabilities.We introduce MED (Measure-Explain-Diagnose), a coarse-to-fine framework that disentangles intrinsic capability changes from… ▽ More Vision tool-use reinforcement learning (RL) can equip vision-language models with visual operators such as crop-and-zoom and achieves strong performance gains, yet it remains unclear whether these gains are driven by improvements in tool use or evolving intrinsic capabilities.We introduce MED (Measure-Explain-Diagnose), a coarse-to-fine framework that disentangles intrinsic capability changes from tool-induced effects, decomposes the tool-induced performance difference into gain and harm terms, and probes the mechanisms driving their evolution. Across checkpoint-level analyses on two VLMs with different tool priors and six benchmarks, we find that improvements are dominated by intrinsic learning, while tool-use RL mainly reduces tool-induced harm (e.g., fewer call-induced errors and weaker tool schema interference) and yields limited progress in tool-based correction of intrinsic failures. Overall, current vision tool-use RL learns to coexist safely with tools rather than master them. △ Less

Submitted 1 February, 2026; originally announced February 2026.

Comments: code: https://github.com/GAIR-NLP/Med

Talk to Me, Not the Slides: A Real-Time Wearable Assistant for Improving Eye Contact in Presentations

Authors: Lingyu Du, Xucong Zhang, Guohao Lan

Abstract: Effective eye contact is a cornerstone of successful public speaking. It strengthens the speaker's credibility and fosters audience engagement. Yet, managing effective eye contact is a skill that demands extensive training and practice, often posing a significant challenge for novice speakers. In this paper, we present SpeakAssis, the first real-time, in-situ wearable system designed to actively a… ▽ More Effective eye contact is a cornerstone of successful public speaking. It strengthens the speaker's credibility and fosters audience engagement. Yet, managing effective eye contact is a skill that demands extensive training and practice, often posing a significant challenge for novice speakers. In this paper, we present SpeakAssis, the first real-time, in-situ wearable system designed to actively assist speakers in maintaining effective eye contact during live presentations. Leveraging a head-mounted eye tracker for gaze and scene view capture, SpeakAssis continuously monitors and analyzes the speaker's gaze distribution across audience and non-audience regions. When ineffective eye-contact patterns are detected, such as insufficient eye contact, or neglect of certain audience segments, SpeakAssis provides timely, context-aware audio prompts via an earphone to guide the speaker's gaze behavior. We evaluate SpeakAssis through a user study involving eight speakers and 24 audience members. Quantitative results show that SpeakAssis increases speakers' eye-contact duration by 62.5% on average and promotes a more balanced distribution of visual attention. Additionally, statistical analysis based on audience surveys reveals that improvements in speaker's eye-contact behavior significantly enhance the audience's perceived engagement and interactivity during presentations. △ Less

Submitted 1 February, 2026; originally announced February 2026.

EFT-CoT: A Multi-Agent Chain-of-Thought Framework for Emotion-Focused Therapy

Authors: Lanqing Du, Yunong Li, YuJie Long, Shihong Chen

Abstract: The use of large language models (LLMs) for Mental Health Question Answering (MHQA) offers a promising way to alleviate shortages in mental health resources. However, prior work has mainly relied on Cognitive Behavioral Therapy (CBT) and predominantly follows a top-down strategy centered on rational cognitive restructuring, providing limited support for embodied experience and primary emotion proc… ▽ More The use of large language models (LLMs) for Mental Health Question Answering (MHQA) offers a promising way to alleviate shortages in mental health resources. However, prior work has mainly relied on Cognitive Behavioral Therapy (CBT) and predominantly follows a top-down strategy centered on rational cognitive restructuring, providing limited support for embodied experience and primary emotion processing. To address this gap, we propose EFT-CoT, a multi-agent chain-of-thought framework grounded in Emotion-Focused Therapy (EFT). EFT-CoT operationalizes intervention as a three-stage workflow: Embodied Perception, Cognitive Exploration, and Narrative Intervention. The framework employs eight specialized agents to model key processes including somatic awareness mapping, adaptive evaluation, core belief extraction, and narrative restructuring. Based on this framework, we construct EFT-Instruct, a high-quality instruction-tuning dataset built from process-level augmentation of about 67,000 real help-seeking texts, and further fine-tune a dedicated model, EFT-LLM. Experiments show that EFT-LLM consistently outperforms strong baselines and human responses in empathic depth and structural professionalism. Ablation studies further verify the contribution of key mechanisms, while white-box auditing demonstrates the consistency and traceability of critical intermediate states. Overall, this work provides a reproducible framework-data-model pipeline for embedding EFT mechanisms into LLM-based mental health support. △ Less

Submitted 8 March, 2026; v1 submitted 25 January, 2026; originally announced January 2026.

Next Generation Active Learning: Mixture of LLMs in the Loop

Authors: Yuanyuan Qi, Xiaohao Yang, Jueqing Lu, Guoxiang Guo, Joanne Enticott, Gang Liu, Lan Du

Abstract: With the rapid advancement and strong generalization capabilities of large language models (LLMs), they have been increasingly incorporated into the active learning pipelines as annotators to reduce annotation costs. However, considering the annotation quality, labels generated by LLMs often fall short of real-world applicability. To address this, we propose a novel active learning framework, Mixt… ▽ More With the rapid advancement and strong generalization capabilities of large language models (LLMs), they have been increasingly incorporated into the active learning pipelines as annotators to reduce annotation costs. However, considering the annotation quality, labels generated by LLMs often fall short of real-world applicability. To address this, we propose a novel active learning framework, Mixture of LLMs in the Loop Active Learning, replacing human annotators with labels generated through a Mixture-of-LLMs-based annotation model, aimed at enhancing LLM-based annotation robustness by aggregating the strengths of multiple LLMs. To further mitigate the impact of the noisy labels, we introduce annotation discrepancy and negative learning to identify the unreliable annotations and enhance learning effectiveness. Extensive experiments demonstrate that our framework achieves performance comparable to human annotation and consistently outperforms single-LLM baselines and other LLM-ensemble-based approaches. Moreover, our framework is built on lightweight LLMs, enabling it to operate fully on local machines in real-world applications. △ Less

Submitted 22 January, 2026; originally announced January 2026.

CD-PIM: A High-Bandwidth and Compute-Efficient LPDDR5-Based PIM for Low-Batch LLM Acceleration on Edge-Device

Authors: Ye Lin, Chao Fang, Xiaoyong Song, Qi Wu, Anying Jiang, Yichuan Bai, Li Du

Abstract: Edge deployment of low-batch large language models (LLMs) faces critical memory bandwidth bottlenecks when executing memory-intensive general matrix-vector multiplications (GEMV) operations. While digital processing-in-memory (PIM) architectures promise to accelerate GEMV operations, existing PIM-equipped edge devices still suffer from three key limitations: limited bandwidth improvement, componen… ▽ More Edge deployment of low-batch large language models (LLMs) faces critical memory bandwidth bottlenecks when executing memory-intensive general matrix-vector multiplications (GEMV) operations. While digital processing-in-memory (PIM) architectures promise to accelerate GEMV operations, existing PIM-equipped edge devices still suffer from three key limitations: limited bandwidth improvement, component under-utilization in mixed workloads, and low compute capacity of computing units (CUs). In this paper, we propose CD-PIM to address these challenges through three key innovations. First, we introduce a high-bandwidth compute-efficient mode (HBCEM) that enhances bandwidth by dividing each bank into four pseudo-banks through segmented global bitlines. Second, we propose a low-batch interleaving mode (LBIM) to improve component utilization by overlapping GEMV operations with GEMM operations. Third, we design a compute-efficient CU that performs enhanced GEMV operations in a pipelined manner by serially feeding weight data into the computing core. Forth, we adopt a column-wise mapping for the key-cache matrix and row-wise mapping for the value-cache matrix, which fully utilizes CU resources. Our evaluation shows that compared to a GPU-only baseline and state-of-the-art PIM designs, our CD-PIM achieves 11.42x and 4.25x speedup on average within a single batch in HBCEM mode, respectively. Moreover, for low-batch sizes, the CD-PIM achieves an average speedup of 1.12x in LBIM compared to HBCEM. △ Less

Submitted 18 January, 2026; originally announced January 2026.

Comments: To appear in 2026 Design, Automation and Test in Europe Conference (DATE 2026)

Optimizing User Profiles via Contextual Bandits for Retrieval-Augmented LLM Personalization

Authors: Linfeng Du, Ye Yuan, Zichen Zhao, Fuyuan Lyu, Emiliano Penaloza, Xiuying Chen, Zipeng Sun, Jikun Kang, Laurent Charlin, Xue Liu, Haolun Wu

Abstract: Large Language Models (LLMs) excel at general-purpose tasks, yet adapting their responses to individual users remains challenging. Retrieval augmentation provides a lightweight alternative to fine-tuning by conditioning LLMs on user history records, and existing approaches typically select these records based on semantic relevance. We argue that relevance serves as an unreliable proxy for utility:… ▽ More Large Language Models (LLMs) excel at general-purpose tasks, yet adapting their responses to individual users remains challenging. Retrieval augmentation provides a lightweight alternative to fine-tuning by conditioning LLMs on user history records, and existing approaches typically select these records based on semantic relevance. We argue that relevance serves as an unreliable proxy for utility: a record may be semantically similar to a query yet fail to improve generation quality or even degrade it due to redundancy or conflicting information. To bridge this gap, we propose PURPLE, a contextual bandit framework that oPtimizes UseR Profiles for Llm pErsonalization. In contrast to a greedy selection of the most relevant records, PURPLE treats profile construction as a set generation process and utilizes a Plackett-Luce ranking model to capture complex inter-record dependencies. By training with dense feedback provided by the likelihood of the reference response, our method aligns retrieval directly with generation quality. Extensive experiments on nine personalization tasks demonstrate that PURPLE consistently outperforms strong heuristic and retrieval-augmented baselines in both effectiveness and efficiency, establishing a principled and scalable solution for optimizing user profiles. △ Less

Submitted 17 January, 2026; originally announced January 2026.

Consolidation or Adaptation? PRISM: Disentangling SFT and RL Data via Gradient Concentration

Authors: Yang Zhao, Yangou Ouyang, Xiao Ding, Hepeng Wang, Bibo Cai, Kai Xiong, Jinglong Gao, Zhouhao Sun, Li Du, Bing Qin, Ting Liu

Abstract: While Hybrid Supervised Fine-Tuning (SFT) followed by Reinforcement Learning (RL) has become the standard paradigm for training LLM agents, effective mechanisms for data allocation between these stages remain largely underexplored. Current data arbitration strategies often rely on surface-level heuristics that fail to diagnose intrinsic learning needs. Since SFT targets pattern consolidation throu… ▽ More While Hybrid Supervised Fine-Tuning (SFT) followed by Reinforcement Learning (RL) has become the standard paradigm for training LLM agents, effective mechanisms for data allocation between these stages remain largely underexplored. Current data arbitration strategies often rely on surface-level heuristics that fail to diagnose intrinsic learning needs. Since SFT targets pattern consolidation through imitation while RL drives structural adaptation via exploration, misaligning data with these functional roles causes severe optimization interference. We propose PRISM, a dynamics-aware framework grounded in Schema Theory that arbitrates data based on its degree of cognitive conflict with the model's existing knowledge. By analyzing the spatial geometric structure of gradients, PRISM identifies data triggering high spatial concentration as high-conflict signals that require RL for structural restructuring. In contrast, data yielding diffuse updates is routed to SFT for efficient consolidation. Extensive experiments on WebShop and ALFWorld demonstrate that PRISM achieves a Pareto improvement, outperforming state-of-the-art hybrid methods while reducing computational costs by up to 3.22$\times$. Our findings suggest that disentangling data based on internal optimization regimes is crucial for scalable and robust agent alignment. △ Less

Submitted 12 April, 2026; v1 submitted 12 January, 2026; originally announced January 2026.

Comments: ACL2026 Main Conference

MAESTRO: Meta-learning Adaptive Estimation of Scalarization Trade-offs for Reward Optimization

Authors: Yang Zhao, Hepeng Wang, Xiao Ding, Yangou Ouyang, Bibo Cai, Kai Xiong, Jinglong Gao, Zhouhao Sun, Li Du, Bing Qin, Ting Liu

Abstract: Group-Relative Policy Optimization (GRPO) has emerged as an efficient paradigm for aligning Large Language Models (LLMs), yet its efficacy is primarily confined to domains with verifiable ground truths. Extending GRPO to open-domain settings remains a critical challenge, as unconstrained generation entails multi-faceted and often conflicting objectives - such as creativity versus factuality - wher… ▽ More Group-Relative Policy Optimization (GRPO) has emerged as an efficient paradigm for aligning Large Language Models (LLMs), yet its efficacy is primarily confined to domains with verifiable ground truths. Extending GRPO to open-domain settings remains a critical challenge, as unconstrained generation entails multi-faceted and often conflicting objectives - such as creativity versus factuality - where rigid, static reward scalarization is inherently suboptimal. To address this, we propose MAESTRO (Meta-learning Adaptive Estimation of Scalarization Trade-offs for Reward Optimization), which introduces a meta-cognitive orchestration layer that treats reward scalarization as a dynamic latent policy, leveraging the model's terminal hidden states as a semantic bottleneck to perceive task-specific priorities. We formulate this as a contextual bandit problem within a bi-level optimization framework, where a lightweight Conductor network co-evolves with the policy by utilizing group-relative advantages as a meta-reward signal. Across seven benchmarks, MAESTRO consistently outperforms single-reward and static multi-objective baselines, while preserving the efficiency advantages of GRPO, and in some settings even reducing redundant generation. △ Less

Submitted 12 April, 2026; v1 submitted 12 January, 2026; originally announced January 2026.

Comments: ACL 2026 Main Conference

Can We Predict Before Executing Machine Learning Agents?

Authors: Jingsheng Zheng, Jintian Zhang, Yujie Luo, Yuren Mao, Yunjun Gao, Lun Du, Huajun Chen, Ningyu Zhang

Abstract: Autonomous machine learning agents have revolutionized scientific discovery, yet they remain constrained by a Generate-Execute-Feedback paradigm. Previous approaches suffer from a severe Execution Bottleneck, as hypothesis evaluation relies strictly on expensive physical execution. To bypass these physical constraints, we internalize execution priors to substitute costly runtime checks with instan… ▽ More Autonomous machine learning agents have revolutionized scientific discovery, yet they remain constrained by a Generate-Execute-Feedback paradigm. Previous approaches suffer from a severe Execution Bottleneck, as hypothesis evaluation relies strictly on expensive physical execution. To bypass these physical constraints, we internalize execution priors to substitute costly runtime checks with instantaneous predictive reasoning, drawing inspiration from World Models. In this work, we formalize the task of Data-centric Solution Preference and construct a comprehensive corpus of 18,438 pairwise comparisons. We demonstrate that LLMs exhibit significant predictive capabilities when primed with a Verified Data Analysis Report, achieving 61.5% accuracy and robust confidence calibration. Finally, we instantiate this framework in FOREAGENT, an agent that employs a Predict-then-Verify loop, achieving a 6x acceleration in convergence while surpassing execution-based baselines by +6%. Our code and dataset are publicly available at https://github.com/zjunlp/predict-before-execute. △ Less

Submitted 7 April, 2026; v1 submitted 9 January, 2026; originally announced January 2026.

Comments: ACL 2026

A Scheduling Framework for Efficient MoE Inference on Edge GPU-NDP Systems

Authors: Qi Wu, Chao Fang, Jiayuan Chen, Ye Lin, Yueqi Zhang, Yichuan Bai, Yuan Du, Li Du

Abstract: Mixture-of-Experts (MoE) models facilitate edge deployment by decoupling model capacity from active computation, yet their large memory footprint drives the need for GPU systems with near-data processing (NDP) capabilities that offload experts to dedicated processing units. However, deploying MoE models on such edge-based GPU-NDP systems faces three critical challenges: 1) severe load imbalance ac… ▽ More Mixture-of-Experts (MoE) models facilitate edge deployment by decoupling model capacity from active computation, yet their large memory footprint drives the need for GPU systems with near-data processing (NDP) capabilities that offload experts to dedicated processing units. However, deploying MoE models on such edge-based GPU-NDP systems faces three critical challenges: 1) severe load imbalance across NDP units due to non-uniform expert selection and expert parallelism, 2) insufficient GPU utilization during expert computation within NDP units, and 3) extensive data pre-profiling necessitated by unpredictable expert activation patterns for pre-fetching. To address these challenges, this paper proposes an efficient inference framework featuring three key optimizations. First, the underexplored tensor parallelism in MoE inference is exploited to partition and compute large expert parameters across multiple NDP units simultaneously towards edge low-batch scenarios. Second, a load-balancing-aware scheduling algorithm distributes expert computations across NDP units and GPU to maximize resource utilization. Third, a dataset-free pre-fetching strategy proactively loads frequently accessed experts to minimize activation delays. Experimental results show that our framework enables GPU-NDP systems to achieve 2.41x on average and up to 2.56x speedup in end-to-end latency compared to state-of-the-art approaches, significantly enhancing MoE inference efficiency in resource-constrained environments. △ Less

Submitted 7 January, 2026; originally announced January 2026.

Comments: To appear in 2026 Design, Automation and Test in Europe Conference (DATE 2026)

Towards Compositional Generalization of LLMs via Skill Taxonomy Guided Data Synthesis

Authors: Yifan Wei, Li Du, Xiaoyan Yu, Yang Feng, Angsheng Li

Abstract: Large Language Models (LLMs) and agent-based systems often struggle with compositional generalization due to a data bottleneck in which complex skill combinations follow a long-tailed, power-law distribution, limiting both instruction-following performance and generalization in agent-centric tasks. To address this challenge, we propose STEPS, a Skill Taxonomy guided Entropy-based Post-training dat… ▽ More Large Language Models (LLMs) and agent-based systems often struggle with compositional generalization due to a data bottleneck in which complex skill combinations follow a long-tailed, power-law distribution, limiting both instruction-following performance and generalization in agent-centric tasks. To address this challenge, we propose STEPS, a Skill Taxonomy guided Entropy-based Post-training data Synthesis framework for generating compositionally challenging data. STEPS explicitly targets compositional generalization by uncovering latent relationships among skills and organizing them into an interpretable, hierarchical skill taxonomy using structural information theory. Building on this taxonomy, we formulate data synthesis as a constrained information maximization problem, selecting skill combinations that maximize marginal structural information within the hierarchy while preserving semantic coherence. Experiments on challenging instruction-following benchmarks show that STEPS outperforms existing data synthesis baselines, while also yielding improved compositional generalization in downstream agent-based evaluations. △ Less

Submitted 7 January, 2026; originally announced January 2026.

Comments: The code and data for our methods and experiments are available at https://github.com/weiyifan1023/STEPS

Let It Flow: Agentic Crafting on Rock and Roll, Building the ROME Model within an Open Agentic Learning Ecosystem

Authors: Weixun Wang, XiaoXiao Xu, Wanhe An, Fangwen Dai, Wei Gao, Yancheng He, Ju Huang, Qiang Ji, Hanqi Jin, Xiaoyang Li, Yang Li, Zhongwen Li, Shirong Lin, Jiashun Liu, Zenan Liu, Tao Luo, Dilxat Muhtar, Yuanbin Qu, Jiaqiang Shi, Qinghui Sun, Yingshui Tan, Hao Tang, Runze Wang, Yi Wang, Zhaoguo Wang , et al. (65 additional authors not shown)

Abstract: Agentic crafting requires LLMs to operate in real-world environments over multiple turns by taking actions, observing outcomes, and iteratively refining artifacts. Despite its importance, the open-source community lacks a principled, end-to-end ecosystem to streamline agent development. We introduce the Agentic Learning Ecosystem (ALE), a foundational infrastructure that optimizes the production p… ▽ More Agentic crafting requires LLMs to operate in real-world environments over multiple turns by taking actions, observing outcomes, and iteratively refining artifacts. Despite its importance, the open-source community lacks a principled, end-to-end ecosystem to streamline agent development. We introduce the Agentic Learning Ecosystem (ALE), a foundational infrastructure that optimizes the production pipeline for agentic model. ALE consists of three components: ROLL, a post-training framework for weight optimization; ROCK, a sandbox environment manager for trajectory generation; and iFlow CLI, an agent framework for efficient context engineering. We release ROME, an open-source agent grounded by ALE and trained on over one million trajectories. Our approach includes data composition protocols for synthesizing complex behaviors and a novel policy optimization algorithm, Interaction-Perceptive Agentic Policy Optimization (IPA), which assigns credit over semantic interaction chunks rather than individual tokens to improve long-horizon training stability. Empirically, we evaluate ROME within a structured setting and introduce Terminal Bench Pro, a benchmark with improved scale and contamination control. ROME demonstrates strong performance across benchmarks like SWE-bench Verified and Terminal Bench, proving the effectiveness of ALE. △ Less

Submitted 11 March, 2026; v1 submitted 31 December, 2025; originally announced December 2025.

Comments: 36 pages, 15 figures

Encyclo-K: Evaluating LLMs with Dynamically Composed Knowledge Statements

Authors: Yiming Liang, Yizhi Li, Yantao Du, Ge Zhang, Jiayi Zhou, Yuchen Wu, Yinzhu Piao, Denghui Cao, Tong Sun, Ziniu Li, Li Du, Bo Lei, Jiaheng Liu, Chenghua Lin, Zhaoxiang Zhang, Wenhao Huang, Jiajun Zhang

Abstract: Benchmarks play a crucial role in tracking the rapid advancement of large language models (LLMs) and identifying their capability boundaries. However, existing benchmarks predominantly curate questions at the question level, suffering from three fundamental limitations: vulnerability to data contamination, restriction to single-knowledge-point assessment, and reliance on costly domain expert annot… ▽ More Benchmarks play a crucial role in tracking the rapid advancement of large language models (LLMs) and identifying their capability boundaries. However, existing benchmarks predominantly curate questions at the question level, suffering from three fundamental limitations: vulnerability to data contamination, restriction to single-knowledge-point assessment, and reliance on costly domain expert annotation. We propose Encyclo-K, a statement-based benchmark that rethinks benchmark construction from the ground up. Our key insight is that knowledge statements, not questions, can serve as the unit of curation, and questions can then be constructed from them. We extract standalone knowledge statements from authoritative textbooks and dynamically compose them into evaluation questions through random sampling at test time. This design directly addresses all three limitations: the combinatorial space is too vast to memorize, and model rankings remain stable across dynamically generated question sets, enabling reliable periodic dataset refresh; each question aggregates 8-10 statements for comprehensive multi-knowledge assessment; annotators only verify formatting compliance without requiring domain expertise, substantially reducing annotation costs. Experiments on over 50 LLMs demonstrate that Encyclo-K poses substantial challenges with strong discriminative power. Even the top-performing OpenAI-GPT-5.1 achieves only 62.07% accuracy, and model performance displays a clear gradient distribution--reasoning models span from 16.04% to 62.07%, while chat models range from 9.71% to 50.40%. These results validate the challenges introduced by dynamic evaluation and multi-statement comprehensive understanding. These findings establish Encyclo-K as a scalable framework for dynamic evaluation of LLMs' comprehensive understanding over multiple fine-grained disciplinary knowledge statements. △ Less

Submitted 6 January, 2026; v1 submitted 31 December, 2025; originally announced December 2025.

IdentityStory: Taming Your Identity-Preserving Generator for Human-Centric Story Generation

Authors: Donghao Zhou, Jingyu Lin, Guibao Shen, Quande Liu, Jialin Gao, Lihao Liu, Lan Du, Cunjian Chen, Chi-Wing Fu, Xiaowei Hu, Pheng-Ann Heng

Abstract: Recent visual generative models enable story generation with consistent characters from text, but human-centric story generation faces additional challenges, such as maintaining detailed and diverse human face consistency and coordinating multiple characters across different images. This paper presents IdentityStory, a framework for human-centric story generation that ensures consistent character… ▽ More Recent visual generative models enable story generation with consistent characters from text, but human-centric story generation faces additional challenges, such as maintaining detailed and diverse human face consistency and coordinating multiple characters across different images. This paper presents IdentityStory, a framework for human-centric story generation that ensures consistent character identity across multiple sequential images. By taming identity-preserving generators, the framework features two key components: Iterative Identity Discovery, which extracts cohesive character identities, and Re-denoising Identity Injection, which re-denoises images to inject identities while preserving desired context. Experiments on the ConsiStory-Human benchmark demonstrate that IdentityStory outperforms existing methods, particularly in face consistency, and supports multi-character combinations. The framework also shows strong potential for applications such as infinite-length story generation and dynamic character composition. △ Less

Submitted 29 December, 2025; originally announced December 2025.

Comments: Accepted by AAAI2026 (Project page: https://correr-zhou.github.io/IdentityStory)

MSC-180: A Benchmark for Automated Formal Theorem Proving from Mathematical Subject Classification

Authors: Sirui Li, Wangyue Lu, Xiaorui Shi, Ke Weng, Haozhe Sun, Minghe Yu, Tiancheng Zhang, Ge Yu, Hengyu Liu, Lun Du

Abstract: Automated Theorem Proving (ATP) represents a core research direction in artificial intelligence for achieving formal reasoning and verification, playing a significant role in advancing machine intelligence. However, current large language model (LLM)-based theorem provers suffer from limitations such as restricted domain coverage and weak generalization in mathematical reasoning. To address these… ▽ More Automated Theorem Proving (ATP) represents a core research direction in artificial intelligence for achieving formal reasoning and verification, playing a significant role in advancing machine intelligence. However, current large language model (LLM)-based theorem provers suffer from limitations such as restricted domain coverage and weak generalization in mathematical reasoning. To address these issues, we propose MSC-180, a benchmark for evaluation based on the MSC2020 mathematical subject classification. It comprises 180 formal verification problems, 3 advanced problems from each of 60 mathematical branches, spanning from undergraduate to graduate levels. Each problem has undergone multiple rounds of verification and refinement by domain experts to ensure formal accuracy. Evaluations of state-of-the-art LLM-based theorem provers under the pass@32 setting reveal that the best model achieves only an 18.89% overall pass rate, with prominent issues including significant domain bias (maximum domain coverage 41.7%) and a difficulty gap (significantly lower pass rates on graduate-level problems). To further quantify performance variability across mathematical domains, we introduce the coefficient of variation (CV) as an evaluation metric. The observed CV values are 4-6 times higher than the statistical high-variability threshold, indicating that the models still rely on pattern matching from training corpora rather than possessing transferable reasoning mechanisms and systematic generalization capabilities. MSC-180, together with its multi-dimensional evaluation framework, provides a discriminative and systematic benchmark for driving the development of next-generation AI systems with genuine mathematical reasoning abilities. △ Less

Submitted 20 December, 2025; originally announced December 2025.

LLaDA2.0: Scaling Up Diffusion Language Models to 100B

Authors: Tiwei Bie, Maosong Cao, Kun Chen, Lun Du, Mingliang Gong, Zhuochen Gong, Yanmei Gu, Jiaqi Hu, Zenan Huang, Zhenzhong Lan, Chengxi Li, Chongxuan Li, Jianguo Li, Zehuan Li, Huabin Liu, Lin Liu, Guoshan Lu, Xiaocheng Lu, Yuxin Ma, Jianfeng Tan, Lanning Wei, Ji-Rong Wen, Yipeng Xing, Xiaolu Zhang, Junbo Zhao , et al. (6 additional authors not shown)

Abstract: This paper presents LLaDA2.0 -- a tuple of discrete diffusion large language models (dLLM) scaling up to 100B total parameters through systematic conversion from auto-regressive (AR) models -- establishing a new paradigm for frontier-scale deployment. Instead of costly training from scratch, LLaDA2.0 upholds knowledge inheritance, progressive adaption and efficiency-aware design principle, and sea… ▽ More This paper presents LLaDA2.0 -- a tuple of discrete diffusion large language models (dLLM) scaling up to 100B total parameters through systematic conversion from auto-regressive (AR) models -- establishing a new paradigm for frontier-scale deployment. Instead of costly training from scratch, LLaDA2.0 upholds knowledge inheritance, progressive adaption and efficiency-aware design principle, and seamless converts a pre-trained AR model into dLLM with a novel 3-phase block-level WSD based training scheme: progressive increasing block-size in block diffusion (warm-up), large-scale full-sequence diffusion (stable) and reverting back to compact-size block diffusion (decay). Along with post-training alignment with SFT and DPO, we obtain LLaDA2.0-mini (16B) and LLaDA2.0-flash (100B), two instruction-tuned Mixture-of-Experts (MoE) variants optimized for practical deployment. By preserving the advantages of parallel decoding, these models deliver superior performance and efficiency at the frontier scale. Both models were open-sourced. △ Less

Submitted 23 December, 2025; v1 submitted 10 December, 2025; originally announced December 2025.

Comments: 19 pages

PrivATE: Differentially Private Average Treatment Effect Estimation for Observational Data

Authors: Quan Yuan, Xiaochen Li, Linkang Du, Min Chen, Mingyang Sun, Yunjun Gao, Shibo He, Jiming Chen, Zhikun Zhang

Abstract: Causal inference plays a crucial role in scientific research across multiple disciplines. Estimating causal effects, particularly the average treatment effect (ATE), from observational data has garnered significant attention. However, computing the ATE from real-world observational data poses substantial privacy risks to users. Differential privacy, which offers strict theoretical guarantees, has… ▽ More Causal inference plays a crucial role in scientific research across multiple disciplines. Estimating causal effects, particularly the average treatment effect (ATE), from observational data has garnered significant attention. However, computing the ATE from real-world observational data poses substantial privacy risks to users. Differential privacy, which offers strict theoretical guarantees, has emerged as a standard approach for privacy-preserving data analysis. However, existing differentially private ATE estimation works rely on specific assumptions, provide limited privacy protection, or fail to offer comprehensive information protection. To this end, we introduce PrivATE, a practical ATE estimation framework that ensures differential privacy. In fact, various scenarios require varying levels of privacy protection. For example, only test scores are generally sensitive information in education evaluation, while all types of medical record data are usually private. To accommodate different privacy requirements, we design two levels (i.e., label-level and sample-level) of privacy protection in PrivATE. By deriving an adaptive matching limit, PrivATE effectively balances noise-induced error and matching error, leading to a more accurate estimate of ATE. Our evaluation validates the effectiveness of PrivATE. PrivATE outperforms the baselines on all datasets and privacy budgets. △ Less

Submitted 16 December, 2025; originally announced December 2025.

Comments: To appear in the NDSS 2026 Symposium, February 2026, San Diego, CA, USA

VICTOR: Dataset Copyright Auditing in Video Recognition Systems

Authors: Quan Yuan, Zhikun Zhang, Linkang Du, Min Chen, Mingyang Sun, Yunjun Gao, Shibo He, Jiming Chen

Abstract: Video recognition systems are increasingly being deployed in daily life, such as content recommendation and security monitoring. To enhance video recognition development, many institutions have released high-quality public datasets with open-source licenses for training advanced models. At the same time, these datasets are also susceptible to misuse and infringement. Dataset copyright auditing is… ▽ More Video recognition systems are increasingly being deployed in daily life, such as content recommendation and security monitoring. To enhance video recognition development, many institutions have released high-quality public datasets with open-source licenses for training advanced models. At the same time, these datasets are also susceptible to misuse and infringement. Dataset copyright auditing is an effective solution to identify such unauthorized use. However, existing dataset copyright solutions primarily focus on the image domain; the complex nature of video data leaves dataset copyright auditing in the video domain unexplored. Specifically, video data introduces an additional temporal dimension, which poses significant challenges to the effectiveness and stealthiness of existing methods. In this paper, we propose VICTOR, the first dataset copyright auditing approach for video recognition systems. We develop a general and stealthy sample modification strategy that enhances the output discrepancy of the target model. By modifying only a small proportion of samples (e.g., 1%), VICTOR amplifies the impact of published modified samples on the prediction behavior of the target models. Then, the difference in the model's behavior for published modified and unpublished original samples can serve as a key basis for dataset auditing. Extensive experiments on multiple models and datasets highlight the superiority of VICTOR. Finally, we show that VICTOR is robust in the presence of several perturbation mechanisms to the training videos or the target models. △ Less

Submitted 16 December, 2025; originally announced December 2025.

Comments: To appear in the NDSS Symposium 2026, February 2026, San Diego, CA, USA

DAPO: Design Structure-Aware Pass Ordering in High-Level Synthesis with Graph Contrastive and Reinforcement Learning

Authors: Jinming Ge, Linfeng Du, Likith Anaparty, Shangkun Li, Tingyuan Liang, Afzal Ahmad, Vivek Chaturvedi, Sharad Sinha, Zhiyao Xie, Jiang Xu, Wei Zhang

Abstract: High-Level Synthesis (HLS) tools are widely adopted in FPGA-based domain-specific accelerator design. However, existing tools rely on fixed optimization strategies inherited from software compilations, limiting their effectiveness. Tailoring optimization strategies to specific designs requires deep semantic understanding, accurate hardware metric estimation, and advanced search algorithms -- capab… ▽ More High-Level Synthesis (HLS) tools are widely adopted in FPGA-based domain-specific accelerator design. However, existing tools rely on fixed optimization strategies inherited from software compilations, limiting their effectiveness. Tailoring optimization strategies to specific designs requires deep semantic understanding, accurate hardware metric estimation, and advanced search algorithms -- capabilities that current approaches lack. We propose DAPO, a design structure-aware pass ordering framework that extracts program semantics from control and data flow graphs, employs contrastive learning to generate rich embeddings, and leverages an analytical model for accurate hardware metric estimation. These components jointly guide a reinforcement learning agent to discover design-specific optimization strategies. Evaluations on classic HLS designs demonstrate that our end-to-end flow delivers a 2.36 speedup over Vitis HLS on average. △ Less

Submitted 12 December, 2025; originally announced December 2025.

Comments: Accepted by DATE 2026

LGAN: An Efficient High-Order Graph Neural Network via the Line Graph Aggregation

Authors: Lin Du, Lu Bai, Jincheng Li, Lixin Cui, Hangyuan Du, Lichi Zhang, Yuting Chen, Zhao Li

Abstract: Graph Neural Networks (GNNs) have emerged as a dominant paradigm for graph classification. Specifically, most existing GNNs mainly rely on the message passing strategy between neighbor nodes, where the expressivity is limited by the 1-dimensional Weisfeiler-Lehman (1-WL) test. Although a number of k-WL-based GNNs have been proposed to overcome this limitation, their computational cost increases ra… ▽ More Graph Neural Networks (GNNs) have emerged as a dominant paradigm for graph classification. Specifically, most existing GNNs mainly rely on the message passing strategy between neighbor nodes, where the expressivity is limited by the 1-dimensional Weisfeiler-Lehman (1-WL) test. Although a number of k-WL-based GNNs have been proposed to overcome this limitation, their computational cost increases rapidly with k, significantly restricting the practical applicability. Moreover, since the k-WL models mainly operate on node tuples, these k-WL-based GNNs cannot retain fine-grained node- or edge-level semantics required by attribution methods (e.g., Integrated Gradients), leading to the less interpretable problem. To overcome the above shortcomings, in this paper, we propose a novel Line Graph Aggregation Network (LGAN), that constructs a line graph from the induced subgraph centered at each node to perform the higher-order aggregation. We theoretically prove that the LGAN not only possesses the greater expressive power than the 2-WL under injective aggregation assumptions, but also has lower time complexity. Empirical evaluations on benchmarks demonstrate that the LGAN outperforms state-of-the-art k-WL-based GNNs, while offering better interpretability. △ Less

Submitted 11 December, 2025; originally announced December 2025.

FLEX: Leveraging FPGA-CPU Synergy for Mixed-Cell-Height Legalization Acceleration

Authors: Xingyu Liu, Jiawei Liang, Linfeng Du, Yipu Zhang, Chaofang Ma, Hanwei Fan, Jiang Xu, Wei Zhang

Abstract: In this work, we present FLEX, an FPGA-CPU accelerator for mixed-cell-height legalization tasks. We address challenges from the following perspectives. First, we optimize the task assignment strategy and perform an efficient task partition between FPGA and CPU to exploit their complementary strengths. Second, a multi-granularity pipelining technique is employed to accelerate the most time-consumin… ▽ More In this work, we present FLEX, an FPGA-CPU accelerator for mixed-cell-height legalization tasks. We address challenges from the following perspectives. First, we optimize the task assignment strategy and perform an efficient task partition between FPGA and CPU to exploit their complementary strengths. Second, a multi-granularity pipelining technique is employed to accelerate the most time-consuming step, finding optimal placement position (FOP), in legalization. At last, we particularly target the computationally intensive cell shifting process in FOP, optimizing the design to align it seamlessly with the multi-granularity pipelining framework for further speedup. Experimental results show that FLEX achieves up to 18.3x and 5.4x speedups compared to state-of-the-art CPU-GPU and multi-threaded CPU legalizers with better scalability, while improving legalization quality by 4% and 1%. △ Less

Submitted 4 December, 2025; originally announced December 2025.

Data Curation Through the Lens of Spectral Dynamics: Static Limits, Dynamic Acceleration, and Practical Oracles

Authors: Yizhou Zhang, Lun Du

Abstract: Large-scale neural models are increasingly trained with data pruning, synthetic data generation, cross-model distillation, reinforcement learning from human feedback (RLHF), and difficulty-based sampling. While several of these data-centric strategies reliably improve training efficiency and downstream performance, others fail to provide meaningful gains -- most notably self-generated synthetic da… ▽ More Large-scale neural models are increasingly trained with data pruning, synthetic data generation, cross-model distillation, reinforcement learning from human feedback (RLHF), and difficulty-based sampling. While several of these data-centric strategies reliably improve training efficiency and downstream performance, others fail to provide meaningful gains -- most notably self-generated synthetic data, which often increases dataset volume without enhancing model capability. We formalize data curation as reweighting the sampling distribution and map its effect onto the eigenstructure of the data-induced operator. Our first main result shows that \textbf{static pruning induces a bounded operator and therefore cannot change the spectral tail exponent}; it provides at most finite-region improvements and cannot alter asymptotic neural scaling. Our second result analyzes \textbf{time-dependent data curation}, showing that an ideal oracle capable of tracking spectral residuals and continuously re-normalizing the tail can provably accelerate learning -- although practical systems can only approximate this behavior. △ Less

Submitted 1 December, 2025; originally announced December 2025.

InnoGym: Benchmarking the Innovation Potential of AI Agents

Authors: Jintian Zhang, Kewei Xu, Jingsheng Zheng, Zhuoyun Yu, Yuqi Zhu, Yujie Luo, Lanning Wei, Shuofei Qiao, Lun Du, Da Zheng, Shumin Deng, Huajun Chen, Ningyu Zhang

Abstract: LLMs and Agents have achieved impressive progress in code generation, mathematical reasoning, and scientific discovery. However, existing benchmarks primarily measure correctness, overlooking the diversity of methods behind solutions. True innovation depends not only on producing correct answers but also on the originality of the approach. We present InnoGym, the first benchmark and framework desi… ▽ More LLMs and Agents have achieved impressive progress in code generation, mathematical reasoning, and scientific discovery. However, existing benchmarks primarily measure correctness, overlooking the diversity of methods behind solutions. True innovation depends not only on producing correct answers but also on the originality of the approach. We present InnoGym, the first benchmark and framework designed to systematically evaluate the innovation potential of AI agents. InnoGym introduces two complementary metrics: performance gain, which measures improvement over the best-known solutions, and novelty, which captures methodological differences from prior approaches. The benchmark includes 18 carefully curated tasks from real-world engineering and scientific domains, each standardized through resource filtering, evaluator validation, and solution collection. In addition, we provide iGym, a unified execution environment for reproducible and long-horizon evaluations. Extensive experiments show that while some agents produce novel approaches, their lack of robustness limits performance gains. These results highlight a key gap between creativity and effectiveness, underscoring the need for benchmarks that evaluate both. △ Less

Submitted 28 February, 2026; v1 submitted 1 December, 2025; originally announced December 2025.

Comments: ICLR 2026

Task-Aware Retrieval Augmentation for Dynamic Recommendation

Authors: Zhen Tao, Xinke Jiang, Qingshuai Feng, Haoyu Zhang, Lun Du, Yuchen Fang, Hao Miao, Bangquan Xie, Qingqiang Sun

Abstract: Dynamic recommendation systems aim to provide personalized suggestions by modeling temporal user-item interactions across time-series behavioral data. Recent studies have leveraged pre-trained dynamic graph neural networks (GNNs) to learn user-item representations over temporal snapshot graphs. However, fine-tuning GNNs on these graphs often results in generalization issues due to temporal discrep… ▽ More Dynamic recommendation systems aim to provide personalized suggestions by modeling temporal user-item interactions across time-series behavioral data. Recent studies have leveraged pre-trained dynamic graph neural networks (GNNs) to learn user-item representations over temporal snapshot graphs. However, fine-tuning GNNs on these graphs often results in generalization issues due to temporal discrepancies between pre-training and fine-tuning stages, limiting the model's ability to capture evolving user preferences. To address this, we propose TarDGR, a task-aware retrieval-augmented framework designed to enhance generalization capability by incorporating task-aware model and retrieval-augmentation. Specifically, TarDGR introduces a Task-Aware Evaluation Mechanism to identify semantically relevant historical subgraphs, enabling the construction of task-specific datasets without manual labeling. It also presents a Graph Transformer-based Task-Aware Model that integrates semantic and structural encodings to assess subgraph relevance. During inference, TarDGR retrieves and fuses task-aware subgraphs with the query subgraph, enriching its representation and mitigating temporal generalization issues. Experiments on multiple large-scale dynamic graph datasets demonstrate that TarDGR consistently outperforms state-of-the-art methods, with extensive empirical evidence underscoring its superior accuracy and generalization capabilities. △ Less

Submitted 16 November, 2025; originally announced November 2025.

Comments: AAAI 2026

DRACO: Co-design for DSP-Efficient Rigid Body Dynamics Accelerator

Authors: Xingyu Liu, Jiawei Liang, Yipu Zhang, Linfeng Du, Chaofang Ma, Hui Yu, Jiang Xu, Wei Zhang

Abstract: We propose a hardware-efficient RBD accelerator based on FPGA, introducing three key innovations. First, we propose a precision-aware quantization framework that reduces DSP demand while preserving motion accuracy. This is also the first study to systematically evaluate quantization impact on robot control and motion for hardware acceleration. Second, we leverage a division deferring optimization… ▽ More We propose a hardware-efficient RBD accelerator based on FPGA, introducing three key innovations. First, we propose a precision-aware quantization framework that reduces DSP demand while preserving motion accuracy. This is also the first study to systematically evaluate quantization impact on robot control and motion for hardware acceleration. Second, we leverage a division deferring optimization in mass matrix inversion algorithm, which decouples reciprocal operations from the longest latency path to improve the performance. Finally, we present an inter-module DSP reuse methodology to improve DSP utilization and save DSP usage. Experiment results show that our work achieves up to 8x throughput improvement and 7.4x latency reduction over state-of-the-art RBD accelerators across various robot types, demonstrating its effectiveness and scalability for high-DOF robotic systems. △ Less

Submitted 22 November, 2025; v1 submitted 11 November, 2025; originally announced November 2025.

Tool Zero: Training Tool-Augmented LLMs via Pure RL from Scratch

Authors: Yirong Zeng, Xiao Ding, Yutai Hou, Yuxian Wang, Li Du, Juyi Dai, Qiuyang Ding, Duyu Tang, Dandan Tu, Weiwen Liu, Bing Qin, Ting Liu

Abstract: Training tool-augmented LLMs has emerged as a promising approach to enhancing language models' capabilities for complex tasks. The current supervised fine-tuning paradigm relies on constructing extensive domain-specific datasets to train models. However, this approach often struggles to generalize effectively to unfamiliar or intricate tool-use scenarios. Recently, reinforcement learning (RL) para… ▽ More Training tool-augmented LLMs has emerged as a promising approach to enhancing language models' capabilities for complex tasks. The current supervised fine-tuning paradigm relies on constructing extensive domain-specific datasets to train models. However, this approach often struggles to generalize effectively to unfamiliar or intricate tool-use scenarios. Recently, reinforcement learning (RL) paradigm can endow LLMs with superior reasoning and generalization abilities. In this work, we address a key question: Can the pure RL be used to effectively elicit a model's intrinsic reasoning capabilities and enhance the tool-agnostic generalization? We propose a dynamic generalization-guided reward design for rule-based RL, which progressively shifts rewards from exploratory to exploitative tool-use patterns. Based on this design, we introduce the Tool-Zero series models. These models are trained to enable LLMs to autonomously utilize general tools by directly scaling up RL from Zero models (i.e., base models without post-training). Experimental results demonstrate that our models achieve over 7% performance improvement compared to both SFT and RL-with-SFT models under the same experimental settings. These gains are consistently replicated across cross-dataset and intra-dataset evaluations, validating the effectiveness and robustness of our methods. △ Less

Submitted 10 November, 2025; v1 submitted 2 November, 2025; originally announced November 2025.

Comments: EMNLP 2025 finding

What Makes AI Research Replicable? Executable Knowledge Graphs as Scientific Knowledge Representations

Authors: Yujie Luo, Zhuoyun Yu, Xuehai Wang, Yuqi Zhu, Ningyu Zhang, Lanning Wei, Lun Du, Da Zheng, Huajun Chen

Abstract: Replicating AI research is a crucial yet challenging task for large language model (LLM) agents. Existing approaches often struggle to generate executable code, primarily due to insufficient background knowledge and the limitations of retrieval-augmented generation (RAG) methods, which fail to capture latent technical details hidden in referenced papers. Furthermore, previous approaches tend to ov… ▽ More Replicating AI research is a crucial yet challenging task for large language model (LLM) agents. Existing approaches often struggle to generate executable code, primarily due to insufficient background knowledge and the limitations of retrieval-augmented generation (RAG) methods, which fail to capture latent technical details hidden in referenced papers. Furthermore, previous approaches tend to overlook valuable implementation-level code signals and lack structured knowledge representations that support multi-granular retrieval and reuse. To overcome these challenges, we propose Executable Knowledge Graphs (xKG), a pluggable, paper-centric knowledge base that automatically integrates code snippets and technical insights extracted from scientific literature. When integrated into three agent frameworks with two different LLMs, xKG shows substantial performance gains (10.9% with o3-mini) on PaperBench, demonstrating its effectiveness as a general and extensible solution for automated AI research replication. Code is available at https://github.com/zjunlp/xKG. △ Less

Submitted 21 January, 2026; v1 submitted 20 October, 2025; originally announced October 2025.

Comments: Work in progress

GRIDAI: Generating and Repairing Intrusion Detection Rules via Collaboration among Multiple LLM-based Agents

Authors: Jiarui Li, Yuhan Chai, Lei Du, Chenyun Duan, Hao Yan, Zhaoquan Gu

Abstract: Rule-based network intrusion detection systems play a crucial role in the real-time detection of Web attacks. However, most existing works primarily focus on automatically generating detection rules for new attacks, often overlooking the relationships between new attacks and existing rules, which leads to significant redundancy within the ever-expanding ruleset. To address this issue, we propose G… ▽ More Rule-based network intrusion detection systems play a crucial role in the real-time detection of Web attacks. However, most existing works primarily focus on automatically generating detection rules for new attacks, often overlooking the relationships between new attacks and existing rules, which leads to significant redundancy within the ever-expanding ruleset. To address this issue, we propose GRIDAI, a novel end-to-end framework for the automated Generation and Repair of Intrusion Detection rules through collaboration among multiple LLM-based agents. Unlike traditional methods, GRIDAI first assesses the nature of incoming attack samples. If the sample represents a new attack type, it is used to generate a new rule. Otherwise, the sample is identified as a variant of an attack already covered by an existing rule and used to repair the rule by updating the corresponding signature, thereby enhancing its generalization capability. Additionally, to mitigate syntactic and semantic errors in rules caused by LLM hallucinations, we incorporate a tool-based real-time validation mechanism and a representative attack sample maintained for each rule, enabling fully automated rule generation and repair. Comprehensive experiments were conducted on a public dataset containing seven types of attacks and a private dataset with 43 attack types. The results demonstrate that GRIDAI accurately identifies the relationships between new attack samples and existing rules, efficiently generates and repairs rules to handle new attacks and variants, and effectively mitigates the impact of LLM hallucinations. △ Less

Submitted 15 October, 2025; originally announced October 2025.

dInfer: An Efficient Inference Framework for Diffusion Language Models

Authors: Yuxin Ma, Lun Du, Lanning Wei, Kun Chen, Qian Xu, Kangyu Wang, Guofeng Feng, Guoshan Lu, Lin Liu, Xiaojing Qi, Xinyuan Zhang, Zhen Tao, Haibo Feng, Ziyun Jiang, Ying Xu, Zenan Huang, Yihong Zhuang, Haokai Xu, Jiaqi Hu, Zhenzhong Lan, Junbo Zhao, Jianguo Li, Da Zheng

Abstract: Diffusion-based large language models (dLLMs) have emerged as a promising alternative to autoregressive (AR) LLMs, leveraging denoising-based generation to enable inherent parallelism. Even more and more open-sourced dLLM models emerge, yet their widespread adoption remains constrained by the lack of a standardized and efficient inference framework. We present dInfer, an efficient and extensible f… ▽ More Diffusion-based large language models (dLLMs) have emerged as a promising alternative to autoregressive (AR) LLMs, leveraging denoising-based generation to enable inherent parallelism. Even more and more open-sourced dLLM models emerge, yet their widespread adoption remains constrained by the lack of a standardized and efficient inference framework. We present dInfer, an efficient and extensible framework for dLLM inference. dInfer decomposes the inference pipeline into four modular components--model, diffusion iteration manager, decoding strategy, and KV-cache manager--and integrates novel algorithms for each component alongside system-level optimizations. Through this combination of algorithmic innovations and system enhancements, dInfer achieves substantial efficiency gains without compromising output quality on LLaDA-MoE. At batch size 1, it surpasses 1,100 tokens per second on HumanEval and averages over 800 tokens per second across six benchmarks on $8\times$ H800 GPUs. Compared to prior systems, dInfer delivers a $10\times$ speedup over Fast-dLLM while maintaining similar model performance. Even compared to the AR model (with a comparable number of activation parameters and performance) QWen2.5-3B, which is highly optimized with the latest vLLM inference engine, dInfer still delivers a $2$-$3\times$ speedup. The implementation of dInfer is open-sourced at https://github.com/inclusionAI/dInfer. △ Less

Submitted 22 October, 2025; v1 submitted 9 October, 2025; originally announced October 2025.

Symmetric Division of Linear Ordinary Differential Operators

Authors: Lixin Du, Manuel Kauers

Abstract: The symmetric product of two ordinary linear differential operators $L_1,L_2$ is an operator whose solution set contains the product $f_1f_2$ of any solution $f_1$ of $L_1$ and any solution $f_2$ of~$L_2$. It is well known how to compute the symmetric product of two given operators $L_1,L_2$. In this paper we consider the corresponding division problem: given a symmetric product $L$ and one of its… ▽ More The symmetric product of two ordinary linear differential operators $L_1,L_2$ is an operator whose solution set contains the product $f_1f_2$ of any solution $f_1$ of $L_1$ and any solution $f_2$ of~$L_2$. It is well known how to compute the symmetric product of two given operators $L_1,L_2$. In this paper we consider the corresponding division problem: given a symmetric product $L$ and one of its factors, what can we say about the other factors? △ Less

Submitted 1 October, 2025; originally announced October 2025.

EvolProver: Advancing Automated Theorem Proving by Evolving Formalized Problems via Symmetry and Difficulty

Authors: Yuchen Tian, Ruiyuan Huang, Xuanwu Wang, Jing Ma, Zengfeng Huang, Ziyang Luo, Hongzhan Lin, Da Zheng, Lun Du

Abstract: Large Language Models (LLMs) for formal theorem proving have shown significant promise, yet they often lack generalizability and are fragile to even minor transformations of problem statements. To address this limitation, we introduce a novel data augmentation pipeline designed to enhance model robustness from two perspectives: symmetry and difficulty. From the symmetry perspective, we propose two… ▽ More Large Language Models (LLMs) for formal theorem proving have shown significant promise, yet they often lack generalizability and are fragile to even minor transformations of problem statements. To address this limitation, we introduce a novel data augmentation pipeline designed to enhance model robustness from two perspectives: symmetry and difficulty. From the symmetry perspective, we propose two complementary methods: EvolAST, an Abstract Syntax Tree (AST) based approach that targets syntactic symmetry to generate semantically equivalent problem variants, and EvolDomain, which leverages LLMs to address semantic symmetry by translating theorems across mathematical domains. From the difficulty perspective, we propose EvolDifficulty, which uses carefully designed evolutionary instructions to guide LLMs in generating new theorems with a wider range of difficulty. We then use the evolved data to train EvolProver, a 7B-parameter non-reasoning theorem prover. EvolProver establishes a new state-of-the-art (SOTA) on FormalMATH-Lite with a 53.8% pass@32 rate, surpassing all models of comparable size, including reasoning-based models. It also sets new SOTA records for non-reasoning models on MiniF2F-Test (69.8% pass@32), Ineq-Comp-Seed (52.2% pass@32), and Ineq-Comp-Transformed (34.0% pass@32). Ablation studies further confirm our data augmentation pipeline's effectiveness across multiple benchmarks. △ Less

Submitted 1 October, 2025; originally announced October 2025.