跳到正文
研踪PaperTrail

论文库

检索计算机领域论文,按研究方向、作者或关键词筛选。

部分采集异常
正在读取账号与研究记录…

全站更新概况

+729新增论文
+575其中 AI 相关
+82已有论文修订
+95新增 AI 导读

arXiv 最近完成扫描:2026-11-21 12:19 · 全站最近入库:2026-11-21 12:19各信源进度 →

统计口径与数据新鲜度

统计 2026-11-20 16:18 — 2026-11-21 16:18(北京时间) · 点击数字查看对应简报;AI 相关包含在新增论文内。

最近完成检查:2026-11-21 12:19最近发现新增版本:2026-11-21 12:19库内论文最新更新时间:2026-11-20 02:59

每分钟刷新已保存的数据。空检查不代表采集异常;新收录可能包含历史论文。AI 导读按新增版本计数。

29,158 已收录论文24,181 AI 相关8,992 有修订112 HF 社区发现

浏览论文库

时间范围跟随排序:版本更新、本站收录或首次发表。最新收录可能包含历史论文。部分信源异常,查看状态 →

找到 44 篇论文 · 按版本更新时间排序上轮扫描起点 2026-11-21 12:17(北京时间)
版本更新 · v2arXiv:2606.10592

Dirichlet-Guided Group Forecasting for Alleviating Over-smoothing in Time Series Forecasting

Xingyu Zhang, Jingyao Wang, Xin Yu, Zeen Song, Jianqi Zhang, Changwen Zheng, Wenwen Qiang

原文摘要

Time series forecasting often suffers from over-smoothing, especially when future dynamics are multi-modal. Forecasts may follow the coarse trend of the observed future, but fail to preserve sharp changes, oscillations, turning points, and regime transitions that define plausible dynamic evolution. In this work, we revisit over-smoothing from the perspective of latent dynamical mode compression: under partial observation and single-realization supervision, multiple plausible future modes can be weakened, merged, or averaged during forecasting. Based on this view, we propose Dirichlet-Guided Group Forecasting (DGF), a mode-preserving forecasting framework that explicitly models multiple mode-conditioned predictive distributions and uncertainty over their selection probabilities. DGF uses a Dirichlet-guided hierarchical sampling mechanism and reward-based optimization to encourage forecasts that are accurate, dynamically consistent, and mode-distinct. Extensive experiments on real-world forecasting benchmarks show that DGF reduces over-smoothing while improving forecasting accuracy, diversity, and dynamical consistency. Our code is publicly available at https://github.com/Zh-XY22/DGF

cs.LG

版本更新 · v3arXiv:2602.06475

Towards Generalizable Reasoning: Group Causal Counterfactual Policy Optimization for LLM Reasoning

Jingyao Wang, Xingyu Zhang, Peizheng Guo, Jiahuan Zhou, Zhenghao Zhang, Changwen Zheng, Chuxiong Sun, Wenwen Qiang

原文摘要

Large language models (LLMs) excel at complex tasks with advances in reasoning capabilities. However, existing reward mechanisms remain tightly coupled to final correctness and pay little attention to the underlying reasoning process: trajectories with sound reasoning but wrong answers receive low credit, while lucky guesses with flawed logic may be highly rewarded, affecting reasoning generalization. From a causal perspective, we interpret multi-candidate reasoning for a fixed question as a family of counterfactual experiments. Building on this, we propose Group Causal Counterfactual Policy Optimization to explicitly train LLMs to learn generalizable reasoning patterns. It proposes an episodic causal counterfactual reward that jointly captures (i) robustness, encouraging the answer distribution induced by a reasoning step to remain stable under counterfactual perturbations; and (ii) effectiveness, enforcing sufficient variability so that the learned reasoning strategy can transfer across questions. We then construct token-level advantages from this reward and optimize the policy, encouraging LLMs to favor reasoning patterns that are process-valid and counterfactually robust. Extensive experiments on diverse benchmarks demonstrate its advantages.

cs.LG

版本更新 · v3arXiv:2507.02844

Visual Contextual Attack: Jailbreaking MLLMs with Image-Driven Context Injection

Ziqi Miao, Yi Ding, Chen Qian, Yongyao Wang, Zaoying Huang, Yuan Xiong, Ken Deng, Lijun Li 等 9 位作者

原文摘要

With the emergence of strong vision language capabilities, multimodal large language models (MLLMs) have demonstrated tremendous potential for real-world applications. However, the security vulnerabilities exhibited by the visual modality pose significant challenges to deploying such models in open-world environments. Recent studies have successfully induced harmful responses from target MLLMs by encoding harmful textual semantics directly into visual inputs. However, in these approaches, the visual modality primarily serves as a trigger for unsafe behavior, often exhibiting semantic ambiguity and lacking grounding in realistic scenarios. In this work, we define a novel setting: vision-centric jailbreak, where visual information serves as a necessary component in constructing a complete and realistic jailbreak context. Building on this setting, we propose the VisCo (Visual Contextual) Attack. VisCo fabricates contextual dialogue using four distinct vision-focused strategies, dynamically generating auxiliary images when necessary to construct a vision-centric jailbreak scenario. To maximize attack effectiveness, it incorporates automatic toxicity obfuscation and semantic refinement to produce a final attack prompt that reliably triggers harmful responses from the target black-box MLLMs. Specifically, VisCo achieves a toxicity score of 4.78 and an Attack Success Rate (ASR) of 85% on MM-SafetyBench against GPT-4o, significantly outperforming the baseline, which achieves a toxicity score of 2.48 and an ASR of 22.2%. Code: https://github.com/Dtc7w3PQ/Visco-Attack.

cs.CV · cs.CL · cs.CR

版本更新 · v3arXiv:2603.28618

Seeing with You: Perception-Reasoning Coevolution for Multimodal Reasoning

Ziqi Miao, Haonan Jia, Lijun Li, Chen Qian, Yuan Xiong, Wenting Yan, Yongyao Wang, Zaoying Huang 等 10 位作者

原文摘要

Reinforcement learning with verifiable rewards (RLVR) has substantially enhanced the reasoning capabilities of multimodal large language models (MLLMs). However, existing RLVR approaches typically rely on outcome-driven optimization that updates both perception and reasoning using a shared reward based solely on the final answer. This shared reward blurs credit assignment, frequently improving reasoning patterns while failing to reliably enhance the accuracy of upstream visual evidence extraction. To address this perception bottleneck, we introduce PRCO (Perception-Reasoning Coevolution), a dual-role RLVR framework with a shared policy. PRCO consists of two cooperative roles: an Observer that generates an evidence caption tailored to the question and a Solver that predicts the final answer based on this caption. Crucially, PRCO employs role-specific reward signals: the Solver is optimized using verifiable outcome rewards on the final answer, while the Observer receives a utility reward derived from the Solver's downstream success. Extensive experiments across eight challenging multimodal reasoning benchmarks demonstrate that PRCO yields consistent improvements across model scales by over 7 points on average accuracy compared to the base model, outperforming prior open-source RL-tuned baselines.

cs.AI

新论文 · v1arXiv:2611.20714

MSAR: Next-Scale Autoregressive Forecasting for Time Series via Modular Multi-Scale Decoupling

Fanda Fan, Kuiye Ding, Liming Mao, Bingsong Wang, Yao Wang, Xiaorui Wang, Ruijie Jian, Zhipeng Liu 等 12 位作者

原文摘要

Time series forecasting underpins critical applications in finance, energy, healthcare, and transportation. Although deep models have achieved strong results, most adopt single-scale modeling or restrict multiscale processing to the input side, causing a misalignment between multiscale inputs and single-scale outputs and limiting predictive power. We introduce the Modular Scale-wise Autoregressive Framework (MSAR), a model-agnostic design that forecasts progressively across multiple temporal resolutions. MSAR offers three advantages: (1) scale-wise aligned modeling, which disentangles heterogeneous temporal patterns by aligning inputs and outputs at each scale; (2) scale-wise autoregression, where coarse-scale predictions guide finer-scale forecasting through hierarchical information flow; and (3) a modular architecture, enabling seamless integration with diverse backbones such as CNNs, MLPs, and Transformers. Extensive experiments across a broad set of datasets and forecasting models demonstrate that MSAR achieves consistent improvements in both accuracy and inference efficiency, validating the effectiveness of scale-aligned autoregression for multiscale time series forecasting. The code is available at: https://github.com/Oryxo/MSAR.

cs.LG

新论文 · v1arXiv:2611.19901

Remember, Replay, Adapt: Learning from Past Experience in Online Vision-and-Language Navigation

Chenhao Ding, Jizhou Han, Lixin Xiu, Muyao Wang, Songlin Dong, Hideki Nakayama

原文摘要

Existing feedback-based methods for online vision-and-language navigation (VLN) primarily learn from current episode outcomes, leaving previously acquired feedback underutilized. Inspired by human episodic memory, we propose R2A, an online VLN adaptation framework that continually reuses successful and failed navigation experiences for policy learning. Specifically, Success-Failure Memory (SFM) retains successful and failed navigation episodes with acquired feedback and retrieves task-relevant experiences based on incoming instructions, enabling historical feedback to guide subsequent navigation. Prior-Anchored Adaptation (PAA) learns from these experiences by reinforcing successful trajectories and suppressing failed ones relative to a frozen pretrained policy, while regularizing behavioral drift. Together, these modules enable continual feedback reuse and stable policy adaptation without source-domain data or expert supervision. Extensive experiments across three benchmarks, REVERIE, R2R, and R2R-CE, and five navigation backbones demonstrate consistent performance gains. On the REVERIE validation unseen split, R2A improves DUET's SR from 46.98% to 69.84% and SPL from 33.73% to 50.98%.

cs.CV · cs.AI

新论文 · v1arXiv:2611.19742

Exploration of Reinforcement Learning in Generative Recommendation Models

Pengfei Zheng, Wuchao Li, Shiyao Wang, Honghui Bao, Hongyi Xie, Jiaxin Deng, Xingmei Wang, Jinghao Zhang 等 14 位作者

原文摘要

Traditional cascaded recommender systems balance accuracy and efficiency over large item spaces, but misaligned stage-wise objectives hinder end-to-end optimization. Generative recommendation unifies these stages in an autoregressive framework, providing a natural interface for reinforcement learning (RL) to optimize the entire generation process. However, directly applying RL often leads to unstable optimization and degraded recommendation quality. We revisit Group Relative Policy Optimization (GRPO) and other ratio-based objectives through a unified logit-gradient view. We identify three issues: (i) clipped surrogates create dead-gradient regions, under-utilizing samples; (ii) negative-advantage updates can become effectively unbounded as importance ratios grow, allowing a small subset of negatives to dominate optimization; and (iii) ratio-based control lacks sensitivity to absolute policy confidence, so updates on high-confidence cases may fail to vanish. Motivated by this diagnosis, we propose \emph{Gradient-Bounded Policy Optimization} (GBPO), a simple objective with stop-gradient normalization that is simultaneously (a) fully gradient-active, (b) sign-symmetrically bounded for both positive and negative updates, and (c) explicitly confidence-weighted under current policy probabilities. This ensures that updates gracefully vanish for $π\!\to\!1$ positives and $π\!\to\!0$ negatives. Across the evaluated settings, GBPO achieves competitive recommendation performance with more stable training than competing RL methods.

cs.LG · cs.AI

新论文 · v1arXiv:2611.17696

Co-Sight Harness: Externalized Control Assets and Deterministic Enforcement for Long-Horizon Business Agents

Luyao Wang, Xijun Zhu, Zhaohui Hao, Xiwen Zhang, Xingyu Wen, Weishan Li, Yuyang Liao, Yanqin Gao 等 14 位作者

原文摘要

Long-horizon business workflows fail in two ways: errors accumulate as a chain of activities lengthens, and the knowledge an activity needs is written in a different vocabulary from the user's request. We study whether such workflows can be made reliable without changing the language model. Our method keeps the backbone fixed and moves the workflow definition, state and domain knowledge into versioned control assets outside the model. A state machine and machine-checkable acceptance gates enforce these assets during execution; an independent auditor grades the resulting trace. The assets are updated only after a regression check and human approval .We evaluate the method on three deployed workflows exceeding 200 steps. On Domain A, a 120-case end-to-end evaluation set shows unattended completion rising from 50% to 100%, retrieval recall from 56% to 91.67%, and end-to-end latency dropping from 440 to 159 minutes, with scenario-analysis recall rising from 45% to 93%. Two other domains were taken over by teams that did not write the assets, one in four days and one in seven days. A controlled ablation separates the two failure sources. The code is available at https://github.com/ZTE-AICloud/Co-Sight/tree/master/Co-Sight%20Harness.

cs.AI

新论文 · v1arXiv:2611.17598

X-Nav Technical Report: Real-World Long-Horizon Indoor-to-Outdoor Navigation on a Humanoid Robot

Jiaxin Huang, Chengxing Lin, Xiaoyu Luo, Sirui Chen, Jingyu Guo, Aolin Li, Kairui Chen, Rui Zhen 等 13 位作者

原文摘要

Long-horizon navigation from indoors to outdoors remains an open problem for humanoid robots: one instruction covers rooms, doors and roads, and the camera moves with the gait. Worken, AGIBOT's measure of productive robot work, depends on robots that reach their workplaces autonomously. We present X-Nav, to our knowledge the first system to take a bipedal humanoid through a language-instructed long-horizon mission from indoors to outdoors. X-Nav covers a pipeline from data to deployment. Its data engine collects navigation data in open-source scenes, in digital twins of deployment sites and on real robots. Its navigation model, MELIOR, takes only RGB images from the robot's built-in camera and comes in two versions: MELIOR-A predicts actions, and MELIOR-N predicts a navigation goal. Its agentic workflow, X-Navarch, turns a brief human instruction into sub-tasks and deploys MELIOR on the robot with explicit target navigation and obstacle avoidance. On the Lingxi X2, X-Nav completes a mission of about a hundred metres from an office lobby to a store in a neighbouring building; LiDAR and odometry are used for target approach and low-level control. The mission is completed with both versions of MELIOR: the robot follows a change of destination and walks through a door from the lobby to the road, and with X-Navarch it also passes static and dynamic obstacles on the route. During deployment, MELIOR is adapted with data collected in a digital twin of the site; the adaptation raises its closed-loop success rate in the twin, and the adapted model is then deployed on the robot at the real site. In addition, MELIOR achieves competitive results on the R2R-CE and RxR-CE benchmarks.

cs.RO · cs.AI · cs.CV

新论文 · v1arXiv:2611.15949

Efficient Influence Minimization via Source-Integrated Sampling

Jingbang Chen, Junyao Wang

原文摘要

Influence minimization by node blocking is a fundamental problem in graph data management that seeks a small set of vertices whose removal minimizes the expected spread from a fixed source set. It supports decisions about where to intervene in large interaction networks to control harmful propagation, such as the spread of rumors and infectious diseases. Under the Independent Cascade (IC) model, selecting blockers requires comparing a large number of candidate sets, and estimating their effects on propagation can be costly. We propose SIG (Source-Integrated Greedy), a sampling framework for efficiently finding high-quality approximate solutions. Our key idea is to group the possible sets of neighbors directly activated by the sources according to their effects on blocker selection. SIG represents these groups in a compact weighted tree and computes their probabilities directly, allowing each sample to account for all source outcomes without enumerating them. We prove that SIG produces unbiased estimates of the lower-bound objective without increasing variance for fixed blocking plans, and establish a high-probability quality guarantee through independent validation. Extensive experiments on real-world networks show that SIG has the lowest mean total running time in all main comparisons, with speedups of up to 583x over the fastest baseline. SIG produces solutions competitive with those of prior methods and achieves the highest influence reduction in multiple settings, including on many of the largest networks, where several baselines exceed the runtime limit. Additional experiments show that SIG maintains consistently low running times as graph size and the blocking budget increase, while delivering stronger blocking solutions at larger budgets.

cs.SI · cs.DS

新论文 · v1arXiv:2611.15834

Where to Generate Touch: Do Image Latents Preserve Material Recognition?

Wanghao Ye, Aarosh Das, Sihan Chen, Yiting Wang, Bowei Tian, Guoheng Sun, Shwai He, Zheyu Shen 等 29 位作者

原文摘要

Generated touch could extend robot learning beyond recorded contacts. A tactile generator needs a latent space, and we choose one by testing whether a material classifier trained on real GelSight presses still works on encode-decode reconstructions from pretrained image models. On ObjectFolder Real, the classifier loses 6.5 to 9.7 points on reconstructions from three 16x discrete tokenizers and 0.3 points on those from the Stable Diffusion VAE (SD-VAE). The 16x tokenizers again lose far more on Touch and Go and on DIGIT object recognition (YCB-Slide). Retraining the classifier on reconstructions, with no new labels, narrows the discrete gap. SD-VAE reconstructions also keep object identity, with 76.7% object-matching accuracy against 78.9% for real touch and 67.0% for MAGVIT-v2. We therefore select SD-VAE latents and show a photo-to-touch generator trained in them. Code will be released upon acceptance.

cs.CV · cs.RO

新论文 · v1arXiv:2611.15617

WaveletLoRA: Frequency-Aware Content-Style Decomposition for Personalized Image Generation

Peiyao Wang, Jiahui Sun, Weining Wang, Jing Liu

原文摘要

Content-style decomposition from a single image is a fundamental challenge in personalized image generation, aiming to separate subject identity from visual style for flexible recomposition. Existing approaches primarily operate in the spatial domain, without explicitly exploiting the distinct frequency characteristics naturally associated with content and style. In this paper, we propose WaveletLoRA, a novel frequency-aware framework for explicit content-style decomposition through wavelet-domain supervision. Specifically, we first apply a multi-level discrete wavelet transform (DWT) to decompose the reference image into low-frequency and high-frequency subbands. We then assign a dedicated LoRA branch to each subband, where the low-frequency branch models global appearance for style acquisition, while the high-frequency branches capture structural details for content preservation. To further improve disentanglement quality, we introduce a dense Mixture-of-Experts aggregation module together with a frequency-domain regularization objective, which reduces style leakage and enhance content fidelity. In addition, we establish WaveBench and develop a VLM-based evaluation protocol that explicitly measures both attribute similarity and attribute separation for disentanglement assessment. Extensive experiments demonstrate that WaveletLoRA consistently outperforms existing methods and produces results that better align with human preference.

cs.CV

版本更新 · v2arXiv:2611.13287

Project Progress Desk: Know Where All Your Claude Code and Codex Projects Stand at a Glance

Wanghao Ye, Yiting Wang, Shwai He, Guoheng Sun, Ziyao Wang, Yexiao He, Zheyu Shen, Zhaoyi Liu 等 25 位作者

原文摘要

Agentic coding tools such as Codex and Claude Code let developers work on many projects in parallel, but keeping track of them still means switching between chats, terminal sessions and providers. We present Project Progress Desk, a local desktop interface that brings overall progress, blockers and pending decisions into one view. Users can see which projects need attention and what remains, and compare earlier approaches before choosing how to continue. Agents report each result in a short receipt; the Desk checks it against the project's checklist and moves the bar, showing earlier estimates in a lighter shade. We describe daily use across 36 projects, including a choice between two implementations based on their recorded outcomes. At the demo, attendees will explore 30 simulated projects, trace completed work to overall progress, and watch a coding agent's new result update the overview.

cs.HC · cs.SE

新论文 · v1arXiv:2611.14098

CUVE: Making Supervised Classification more Accessible through Visualization

Sanchitha Kuppusami, Xiyao Wang, Miguel Nacenta

原文摘要

We present CUVE, an interface that enables people to build classifiers by generating visualizations that are interpretable by both humans and learning algorithms (the Visualization as Intermediate Representation approach). This UI enables users to create supervised classifiers and to visually investigate how to improve them and how they carry out their function. A preliminary study investigates the usability of our prototype and compares it with a traditional text-based programming interface.

cs.HC · cs.LG

新论文 · v1arXiv:2611.13931

TAPCO-G: LLM-Guided Design-Adaptive Timing and Power Co-Optimization with Graph-Based Physical Priors

Wuqian Tang, Zixiao Wang, Che-Chien Lin, Fong-Hua Fu, Laura Angélica Shion-Korsak, Xinyun Zhang, Bei Yu, Chun-Yao Wang

原文摘要

Post-placement timing and power co-optimization is critical to meeting setup timing targets and reducing power within placement constraints. A cell replacement can improve timing along one path while increasing power or reducing slack on others. Conventional optimizers follow fixed rules for scheduling and accepting repairs, limiting their ability to address design-specific bottlenecks. A large language model (LLM) can propose changes to these rules, but on its own it cannot reliably estimate how a cell replacement affects endpoint slack and power. TAPCO-G combines LLM-guided optimizer adaptation with graph-based physical priors to evaluate candidate cell replacements. The outer loop tunes tool settings using physical-design feedback and revises executable repair policies when tuning stalls. The inner loop anchors delay estimates to static timing analysis snapshots, accounts for upstream driver-load changes, and propagates delay shifts to timing endpoints. It combines these estimates with coarse-to-fine power screening to prioritize candidates for physical trials. TAPCO-G produced placement-legal, logic-equivalent implementations and outscored CERBERUS on all eight ASAP7 benchmarks from the MLCAD 2026 Contest, with a mean quality score gain of 5.516 points, and scored 21.648 points higher than TAPCO on average. These results demonstrate the effectiveness of the proposed co-optimization framework.

cs.AR · cs.AI

版本更新 · v4arXiv:2606.24548

Are Text-to-Image Models Inductivist Turkeys? A Counterfactual Benchmark for Causal Reasoning

Jiayi Lei, Yuandong Pu, Xingyu Han, Rongpeng Zhu, Jing Xu, Jinyao Wang, Zijian Zhou, Bin Fu 等 11 位作者

原文摘要

Text-to-image (T2I) generation models have achieved remarkable progress in producing visually realistic images from natural language prompts. Yet it remains unclear whether their success reflects genuine causal understanding or sophisticated pattern matching over visual-textual correlations. Inspired by Russell's inductivist turkey, we introduce Counterfactual-World (CF-World), a counterfactual benchmark designed to investigate whether text-to-image models can generate images under rules that systematically contradict real-world priors. CF-World organizes each scenario into three progressive levels: factual generation under ordinary world knowledge, explicit counterfactual generation with direct visual instructions, and implicit counterfactual generation requiring causal deduction from altered rules. We evaluate open-source and closed-source text-to-image models using CF-Eval, a scoring pipeline based on vision-language models. We report the scores at each prompt level as the primary results and use Prior Resistance Rate (PRR) and Reasoning Retention Rate (RRR) as auxiliary descriptive indicators. All evaluated models score lower on counterfactual prompts than on factual prompts. Diagnostic experiments reveal difficulties in following altered rules and recombining rarely co-occurring visual attributes. The gains from rephrasing object and attribute names are consistent with a role for learned visual priors in counterfactual generation. Our benchmark and evaluation code are publicly available at https://jylei16.github.io/CF-World.github.io/.

cs.CV

版本更新 · v2arXiv:2607.26515

HiFloat4 Format for End-To-End Reinforcement Learning Post-Training of Large Language Models

Hei Yi Mak, Shadan Golestan, Hoang Le, Mehran Taghian Jazi, Yunke Peng, Yaoyuan Wang, Yao Wang, Junsong Wang 等 13 位作者

原文摘要

We present, to our knowledge, the first end-to-end FP4 RL post-training, in which both the rollout and training policies, including their forward and backward passes, operate at 4-bit precision. A systematic study reveals that the dominant source of degradation in FP4 RL is not training-side quantization error but rollout activation quantization: outliers stretch the dynamic range so far that a large number of activation values underflow to zero under FP4. Counterintuitively, restoring the training policy to higher precision while keeping the rollout in FP4 makes accuracy worse than full FP4 baseline, exposing rollout--training mismatch as the principal failure mode and ruling out standard pretraining-style fixes. We address this with Rollout Residual Quantization (Rollout-ResQ): a single residual correction term constrained to a hardware-friendly sparsity pattern, added only to the FP4 rollout matmul -- a lightweight correction that recovers most of the precision lost to outlier-driven underflow without inflating the rollout's compute footprint. On Qwen2.5-3B, Rollout-ResQ paired with the HiFloat4 (HiF4) format -- whose three-level hierarchical scaling preserves resolution under FP4's tight 4-bit budget -- closes the accuracy gap to BF16 from 3.59% to 1.39%, bringing fully quantized FP4 RL within close reach of full precision. Applied to the open-standard MXFP4, the same recipe narrows the gap from 13.6% to 5.3%, revealing that FP4 format choice is a key factor that determines the ceiling on recoverable accuracy. The experimental results on Qwen2.5-Math-7B further confirm the recovery benefits of Rollout-ResQ with HiF4. Together, these results establish HiF4 as the enabling format for end-to-end FP4 RL post-training, and Rollout-ResQ as the activation-side mechanism that makes the gap to BF16 closable.

cs.LG · cs.AI

版本更新 · v4arXiv:2602.09843

Kelix Technical Report

Boyang Ding, Chenglong Chu, Dunju Zang, Han Li, Jiangxia Cao, Kun Gai, Muhao Wei, Ruiming Tang 等 31 位作者

原文摘要

Autoregressive large language models (LLMs) scale well by expressing diverse tasks as sequences of discrete natural-language tokens and training with next-token prediction, which unifies comprehension and generation under self-supervision. Extending this paradigm to multimodal data requires a shared, discrete representation across modalities. However, most vision-language models (VLMs) still rely on a hybrid interface: discrete text tokens paired with continuous Vision Transformer (ViT) features. Because supervision is largely text-driven, these models are often biased toward understanding and cannot fully leverage large-scale self-supervised learning on non-text data. Recent work has explored discrete visual tokenization to enable fully autoregressive multimodal modeling, showing promising progress toward unified understanding and generation. Yet existing discrete vision tokens frequently lose information due to limited code capacity, resulting in noticeably weaker understanding than continuous-feature VLMs. We present Kelix, a fully discrete autoregressive unified model that closes the understanding gap between discrete and continuous visual representations.

cs.CV

版本更新 · v3arXiv:2610.37161HF 社区推荐

WorkFactory: Turning Professional Intents into Executable Work for Scalable Agent Training

Ziyang Zeng, Tingyang Chen, Jia Deng, Qiyao Wang, Shuo Tang, Yike Yang, Tao Ren, Wayne Xin Zhao 等 13 位作者

原文摘要

Scaling supervised fine-tuning for working agents requires not only broad coverage of professional tasks, but also expert demonstrations that faithfully carry out professional workflows to completion. However, missing or inconsistent task information can disrupt expert demonstrations, weakening the resulting supervision. We introduce WorkFactory, a scalable task-construction framework that turns professional intents into executable task environments through hybrid intent sourcing, semantic workflow planning, and execution-based repair. Starting from these intents, WorkFactory constructs a Semantic Workflow Graph that explicitly represents information requirements, dependencies, and locations within the task environment. The graph guides the instantiation of task environments, while an adversarial execution-refinement loop exposes missing or inconsistent information and steers targeted repairs that preserve the original intent. Using WorkFactory, we construct 112,686 tasks spanning 612 occupations and train the WorkPilot family at 4B, 27B, and 35B-A3B scales on the resulting teacher trajectories. All three models improve over their respective base models across four professional-work benchmarks, with WorkPilot-35B-A3B rising from 879 to 1565 Elo on GDPval-AA v2.1. Scaling experiments with up to 110k tasks further show that larger models benefit more from additional task data but exhibit diminishing returns at smaller data scales, while WorkPilot's long-horizon execution evolves from verbose trajectories toward more efficient task completion. Controlled component studies quantify the negative effects of task-environment defects on downstream performance and show how semantic workflow planning and execution-based repair yield downstream gains through improved task-environment quality. The subset of data are available at https://github.com/IQuestLab/WorkFactory.

cs.AI

版本更新 · v3arXiv:2604.07383

SCOT: Soft Correspondence via Optimal Transport for Multi-Source Cross-City Transfer

Yuyao Wang, Min Yang, Meng Chen, Weiming Huang, Yilong Yin, Yongshun Gong

原文摘要

Cross-city transfer uses labeled data from well-instrumented cities to improve prediction in label-scarce ones. The task is difficult when cities have incompatible region partitions and no ground-truth correspondence. Existing methods either rely on heuristic anchors, which can create hubness, or align global distributions, which can over-mix heterogeneous regions. We propose SCOT (Soft Correspondence via Optimal Transport), an optimal transport-based framework for learning explicit, mass-controlled soft correspondences between unequal region sets. SCOT combines an optimal transport-weighted contrastive loss, a one-sided cycle regularizer, and a shared prototype hub for multi-source transfer. Across real-world cities and tasks, SCOT consistently outperforms strong baselines, reducing MAE and MAPE by 5-50%. Additional analyses show that the learned correspondences are selective, region-specific, and robust to heterogeneous settings.

cs.LG

新论文 · v1arXiv:2611.10798

RTLREX: LLM-Guided Framework for Exact High-Level RTL Recovery

Wuqian Tang, Fangzhou Liu, Ruijie Wang, TingTing Hwang, Bei Yu, Chun-Yao Wang

原文摘要

Hardware designs may be available only as gate-level netlists or input-output descriptions when the original register-transfer-level (RTL) design and documentation are not accessible. Low-level representations obscure word-level computation and its finite-width behavior. Recovering high-level RTL from such representations therefore helps engineers understand a circuit's functionality and enables RTL rewriting and resynthesis. A plausible function model is also insufficient unless the generated RTL is equivalent to the original circuit for all input combinations. In this paper, we present RTLREX, a large language model (LLM)-guided framework that coordinates domain-specific tools for function recovery, RTL generation, independent verification, and cost-guided rewriting. Independent verification establishes equivalence between the recovered RTL and the original circuit. The LLM uses tool feedback to select further analyses, revise candidate RTL, and propose syntax-cost reductions. Experiments on 30 benchmarks from Problem A of the ICCAD 2022 CAD Contest show that RTLREX recovers equivalent high-level RTL for all benchmarks, surpassing the state-of-the-art method WolFEx in recovery coverage. On benchmarks recovered by both methods, RTLREX reduces mean syntax cost by 20.5% relative to WolFEx. On complete truth-table benchmarks from the IWLS 2026 Contest, our framework reduces mean syntax cost by 95.7% relative to an LLM agent baseline on the benchmarks validated by both methods.

cs.AR · cs.AI

版本更新 · v2arXiv:2611.00403

MXCode: Vector Quantization for MX-Native Inference at Scale

Gaurav Kolhe, Cansu Demirkiran, Shubhi Tyagi, Sulaiman Sadiq, Boyao Wang, Richard Wang, Maximiliana Behnke, Shreyas Labhsetwar 等 11 位作者

原文摘要

Memory capacity and bandwidth are major bottlenecks in LLM inference, motivating weight compression beyond conventional 4-bit quantization. Vector quantization (VQ) enables approximately 2-bit weight storage but often relies on codebooks incompatible with native FP4 computation. We introduce MXCode (MX-Native Codebook Quantization), a framework bridging ultra-low-bit VQ compression with MicroXcaling (MX)-native FP4 computation. MXCode proposes three strategies: PostScale, which quantizes learned codebooks to MXFP4; PreScale, which extracts MX scales before codebook learning; and PreScale+, which further compresses these scales using VQ. These strategies explore tradeoffs among quantization quality, scale overhead, and codebook granularity. All reconstruct MX-native W4A4 operands without materializing intermediate high-precision weights. We evaluate MXCode on Qwen3 (235B), DeepSeek-R1 (671B), and GLM-5.2 (753B) using the Microsoft MX Emulation Library. MXCode achieves approximately 2 bits per weight with competitive, workload-dependent model quality. On GB300 hardware, experiments with Qwen3 (235B) show approximately 50% lower model-load memory than Native FP4 while maintaining near-Native-FP4 serving throughput. Under memory-constrained prefix reuse, P90 TTFT improves by up to 94.5%. These results demonstrate software-based reconstruction on existing FP4 hardware and inform the development of future MX-native inference systems.

cs.LG

版本更新 · v3arXiv:2605.01507

MILD: Mediator Agentic System with Bidirectional Perception and Multi-Layered Alignment for Human-Vehicle Collaboration

Jiyao Wang, Yunbiao Wang, Yubo Jiao, Xiao Yang, Dengbo He, Sasan Jafarnejad, Luis Miranda-Moreno, Raphael Frank 等 9 位作者

原文摘要

Prior studies report that partial driving automation can increase the cognitive demands on human drivers. This effect largely arises from human drivers' lack of transparent insight into the vehicle's intentions and decision logic, as well as from automated systems' limited awareness of the driver's dynamic state and preferences. This bidirectional misalignment undermines shared situational awareness and exacerbates coordination failures in human-vehicle interaction. To address these limitations, we argue for a paradigm shift that elevates the human role from passive supervisor to active manager. We introduce the Mediator-in-the-Loop-Driving system, based on an agentic system architecture to facilitate synergistic human-vehicle collaboration. MILD integrates a perception agent for joint in-cabin and out-of-cabin understanding with a lightweight strategy agent that generates compliant and explainable action suggestions. To ensure these strategies are strictly aligned with safety regulations and human values, we develop Evidence- and Constraint-weighted Policy Optimization. ECPO leverages automatic validators to steer the agent toward behaviors that are not only accurate but also structurally complete, substantiated by evidence, and less prone to constraint violations. Furthermore, a retrieval-augmented generation module dynamically incorporates constraints from traffic regulations, speed recommendations, and driver preferences into the decision loop. Experiments across three open datasets demonstrate that MILD consistently outperforms baselines in both perception accuracy and strategy quality under auditable offline metrics, and yields higher human-rated policy adequacy, comfort, and explanation than baselines.

cs.AI

版本更新 · v5arXiv:2510.23254

Optimal In-context Adaptivity and Distributional Robustness of Transformers

Tianyi Ma, Tengyao Wang, Richard J. Samworth

原文摘要

We study in-context learning problems where a Transformer is pretrained on tasks drawn from a mixture distribution $π=\sum_{α\in\mathcal{A}} λ_α π_α$, called the pretraining prior, in which each mixture component $π_α$ is a distribution on tasks of a specific difficulty level indexed by $α$. Our goal is to understand the performance of the pretrained Transformer when evaluated on a different test distribution $μ$, consisting of tasks of fixed difficulty $β\in\mathcal{A}$, and with potential distribution shift relative to $π_β$, subject to the chi-squared divergence $χ^2(μ,π_β)$ being at most $κ$. In particular, we consider nonparametric regression problems with random smoothness, and multi-index models with both random smoothness and random effective dimension. We prove that a large Transformer pretrained on sufficient data achieves the optimal rate of convergence corresponding to the difficulty level $β$, uniformly over test distributions $μ$ in the chi-squared divergence ball. Thus, the pretrained Transformer is able to achieve faster rates of convergence on easier tasks and is robust to distribution shift at test time. Finally, we prove that even if an estimator had access to the test distribution $μ$, the convergence rate of its expected risk over $μ$ could not be faster than that of our pretrained Transformers, thereby providing a more appropriate optimality guarantee than minimax lower bounds.

stat.ML · cs.LG · math.ST

版本更新 · v2arXiv:2611.04376

Aligning Training and Rollout Policies for End-to-End FP4 RL on Mixture-of-Experts LLMs

Hei Yi Mak, Hoang Le, Mehran Taghian Jazi, Yunke Peng, Huan Lin, Zhengzhi Tang, Zhexu Liu, Yao Wang 等 11 位作者

原文摘要

In this work, we study end-to-end FP4 reinforcement learning (RL) post-training for Mixture-of-Experts (MoE) large language models. Recent work on FP4 RL has introduced effective recovery recipes that restore accuracy under aggressive low-precision quantization for dense models, but we find that the same approach yields no improvement over the naive FP4 baseline in the MoE setting. We conduct an in-depth study to understand why. Our analysis identifies the root cause in the routing mechanism unique to MoE: because the router makes discrete expert-selection decisions, training-inference policy alignment becomes acutely sensitive to whether the same recovery technique is applied symmetrically on the training and inference sides. When an asymmetric recipe is used-including the dense-model approach, which corrects only the rollout-training-inference alignment deteriorates sharply, as measured by the Pearson correlation coefficient between training and inference probabilities, and router selection matching rates. Building on this finding, we propose MoEResQ, a quantization-recovery recipe for MoE RL post-training that enforces symmetry between the training- and inference-side corrections. MoEResQ restores training-inference alignment and substantially closes the accuracy gap to full precision, establishing symmetric recovery as the key principle in FP4 RL for MoE models.

cs.LG · cs.AI

版本更新 · v2arXiv:2508.01128

When Reviews Are Scarce: Structure-Aware Review Imputation for Recommendation in Realistic Sparse Review Settings

Leyao Wang, Xutao Mao, Qinwen Ge, Xuhui Zhan, Yuying Zhao, Bo Ni, Ryan A. Rossi, Nesreen K. Ahmed 等 9 位作者

原文摘要

Textual reviews enrich recommender systems with fine-grained preference signals and enhanced explainability. However, in real-world scenarios, users rarely leave reviews, resulting in severe sparsity that undermines the effectiveness of review-aware recommenders under partial observability. A natural remedy is to synthesize the missing reviews to restore the textual signal these recommenders rely on. Yet, conventional imputation techniques, such as matrix completion and LLM-based augmentation, either lose contextualized semantics by embedding texts into vectors or overlook structural dependencies on the user-item graph. To address these shortcomings, we propose TWISTER (ToWards Imputation on Sparsity with Textual Edge Graph Representation), a unified framework for sparse-text recommendation that synthesizes missing reviews by jointly modeling semantic and structural signals. Specifically, we represent user-item interactions as a Textual-Edge Graph (TEG), treating reviews as edge attributes. To capture relational context, we construct line-graph views and employ an LLM as a graph-aware aggregator. For each interaction lacking a textual review, our model aggregates natural-language representations from neighboring interactions to generate a coherent and personalized review. Experiments on multiple datasets show that TWISTER consistently outperforms traditional numeric, graph-based, and LLM baselines, delivering higher-quality imputed reviews and, more importantly, improved downstream recommendation. In summary, TWISTER is a drop-in, structure-aware imputer that recovers the textual signal that review-aware recommenders need when reviews are scarce.

cs.IR · cs.AI · cs.LG

新论文 · v1arXiv:2611.04673

Right Tool, Wrong Parameters: Defending LLM Agents Against Parameter Poisoning

Yao Wang, Zehua Ma, Weiming Zhang, Han Fang

原文摘要

Indirect Prompt Injection (IPI) poses a critical threat to LLM agents. Existing defenses largely focus on whether an injection induces a new tool invocation or unintended action. In this paper, we study a different attack surface, parameter poisoning, where the attacker preserves the intended action but manipulates one or more tool arguments, such as a recipient or other task-critical value. Such attacks are difficult to detect because no new action is introduced, while parameter changes themselves are often legitimate parts of the user's request. Our key insight is that the legitimacy of a parameter change should be determined by the trust of the information flow that produces it under the user query. The query induces a task-specific trust hierarchy over available information fields: some are trusted to determine particular parameters, while others should have little or no influence. Based on this insight, we propose AuthorityFlow, a three-stage parameter-level defense. First, it interprets the user query to construct a task-specific trust hierarchy over the information fields available to the agent. Second, for each tool argument, it traces the information flow that produces its current value and attributes the value to its supporting field. Finally, it accepts the value only when that field is trusted to determine the parameter under the current task, allowing high-trust changes while flagging low-trust influence as potential parameter poisoning. We further introduce FactAuthBench for evaluating parameter poisoning in LLM agents. Without defense, the attack success rate reaches 93.06%. AuthorityFlow reduces it to 9.03%, while retaining 82.64% benign task utility compared with 90.28% without defense. Our results show that effective IPI defense requires reasoning not only about what action an agent takes, but also about which information is trusted to determine its parameters.

cs.CR

新论文 · v1arXiv:2611.03410

OpenZilo: An Open Smart-Ring Platform for Human-Centric Sensing and Interaction with Robots

Qianggang Ding, Xingyao Wang, Rui Feng, Chi Qin, Weijian Lin, Lizhi Ding, Jiankai Tang, Yuntao Wang 等 10 位作者

原文摘要

Wearable robot interfaces must connect user input to timely physical action. We present the proposed platform, an open smart-ring platform for studying this connection across task stages. The ring integrates a microphone, an inertial measurement unit, photoplethysmography, and skin-temperature sensing. Its software architecture separates acquisition, event interpretation, interaction policy, and robot execution. We characterize seven operating modes and evaluate speech and gesture control with 10 participants. Across five commands and 250 trials per modality, speech and gestures achieved task success rates of 90.4% and 88.8%, respectively. Recognition accuracy exceeded task success for both modalities. Gaussian hidden Markov models achieved higher accuracy than a support vector machine across three vocabulary sizes, with longer scoring times. Following a shared spoken task start, gesture-based adjustments had shorter median response times than speech. Whole-task success was 70.0% with gestures and 64.0% with speech. We further examine physiological-event-triggered robot interaction using simulated event notifications. Mean notification-receipt-to-arrival times were 11.71 s for the robot and 9.20 s for an on-site person. These descriptive findings support evaluating wearable robot interfaces through both recognition and physical outcomes, with timing measures appropriate to each task stage.

cs.RO · cs.HC

版本更新 · v2arXiv:2610.35013HF 社区推荐

Multiscale Policy Optimization for Long-Horizon LLM Agents

Tao Ren, Hui Yang, Jia Deng, Tingyang Chen, Qiyao Wang, Shuo Tang, Ziyang Zeng, Yuhua Jiang 等 19 位作者

原文摘要

Long-horizon LLM agents interact with their environments through complete assistant turns while generating each action token by token. We observe that critic values vary primarily across turns, motivating optimization anchored at turns with finer credit allocation among tokens. We propose Multiscale Policy Optimization (MPO), combining Multiscale Advantage Estimation (MAE) and turn-level importance sampling to align credit assignment and policy updates at the turn scale. Extensive experiments across search, tool-use, and automation tasks with 4B and 30B-A3B models demonstrate consistent gains over strong RL baselines.

cs.AI

新论文 · v1arXiv:2611.02698

GeoGuesser Bench: Benchmarking Vision-Language Agents on Interactive Geolocation

Huanzhi Mao, Yichuan Wang, Michael Koo, Wenyuan Huang, Huy Doan, Jinghan Xia, Ziyao Wang, Joseph E. Gonzalez

原文摘要

Vision-language model (VLM) agents that act in the open world need two abilities: world knowledge to interpret what they see, and exploration to find what they cannot yet see. We study how agents combine the two through interactive visual geolocation. GeoGuesser Bench contains 282 tasks in an offline Street View environment where agents rotate, zoom, and move between real panoramas. Its three task types shift the balance from knowledge to exploration: agents (1) report the coordinates of their starting location, (2) name the enclosing street, city, state, or country, or (3) find, read, and count nearby objects. The coordinate tasks in (1) also include 42 real GeoGuessr World Championship rounds, whose expert guesses provide a human reference. We evaluate 23 VLMs: the best accuracy is 62.5% on place naming and 35.0% on object finding, and the best agent averages 4352.4 GeoGuessr points on the championship rounds, against 4770.3 for the better expert in each round. Beyond these headline numbers, world knowledge and exploration are only weakly correlated across models: agents that localize well do not necessarily explore well. To understand where agents fail, we categorize their errors into vision, navigation, and tool-use failures. Further ablations show that the strongest agents, such as Claude Opus 4.8, still perform well with limited views, while weaker agents degrade; customizing the environment thus adjusts task difficulty. The environment's verifiable rewards and restorable state also support training, and our Planner-Worker-Compressor harness cuts planner tokens by 57-61% by compacting the interaction history. We hope GeoGuesser Bench and its environment serve as a high-quality foundation for simulating real-world exploration at scale.

cs.AI · cs.LG