Shuo Lu

Yiyou Sun, Xinyang Han, Weichen Zhang et al.

Recent AI systems have achieved strong results on a wide range of benchmarks, yet these gains have not translated into economically meaningful deployment across many professional domains. We argue that this gap is largely an evaluation problem: widely used benchmarks lack sustained performance measurement on real and economically valuable workflows. This paper introduces Agents' Last Exam (ALE), a benchmark designed to evaluate AI agents on long-horizon, economically valuable, real-world tasks with verifiable outcomes. Developed in collaboration with 250+ industry experts, ALE covers non-physical industries defined with reference to O*NET / SOC 2018 (the U.S. federal occupational taxonomy). It is organized around a task taxonomy with 55 subfields grouped into 13 industry clusters covering 1K+ tasks. Current results show that the hardest tier remains far from saturated: across mainstream harness and backbone configurations, the average full pass rate is 2.6%. ALE is designed as a living benchmark: its task pool grows continuously as new workflows and industries are onboarded. More broadly, ALE is intended not merely as another leaderboard, but as an instrument for closing the gap between benchmark success and GDP-relevant impact.

Shuo Lu, Yinuo Xu, Kecheng Yu et al.

Large language models (LLMs) are increasingly asked not only to write static interfaces, but to construct executable interactive worlds from natural language. Browser-native 3D, commonly built with Three.js, is a natural next frontier: generated programs must integrate assets, obey spatial and physical constraints, and keep user-facing controls synchronized with hidden runtime state. Existing web-generation benchmarks and evaluators, however, largely observe only pixels or DOM nodes, while the mechanics of a Three.js world unfold inside an opaque <canvas>. We introduce WorldCoder-Bench, a benchmark for autonomous, physically grounded 3D world synthesis. WorldCoder-Bench contains 2,026 expert-curated tasks across Simulation, Rendering, and Application scenarios, with optional .glb assets and hidden behavioral contracts. We further propose StateProbe, an execution-based protocol that probes generated programs in a sandboxed browser and verifies hidden, mutation-hardened contracts over runtime states and transitions. Beyond verification coverage, we report Return on Automation and Time Efficiency Multiplier to measure correctness-adjusted cost and time savings. Across nine frontier models, the best system reaches only 27.8% verification coverage on WorldCoder-Core and 19.9% on WorldCoder-Robust, with failures dominated by state-schema drift and broken interaction chains rather than missing scene elements. Utility metrics further show that cheap or fast models can still provide substantial value on easier domains. WorldCoder-Bench is available at https://anonymous.4open.science/r/WorldCoder-Bench/.

Shuo Lu, Yingsheng Wang, Lijun Sheng et al.

Out-of-distribution (OOD) detection aims to detect test samples outside the training category space, which is an essential component in building reliable machine learning systems. Existing reviews on OOD detection primarily focus on method taxonomy, surveying the field by categorizing various approaches. However, many recent works concentrate on non-traditional OOD detection scenarios, such as test-time adaptation, multi-modal data sources and other novel contexts. In this survey, we uniquely review recent advances in OOD detection from the task-oriented perspective for the first time. According to the user's access to the model, that is, whether the OOD detection method is allowed to modify or retrain the model, we classify the methods as training-driven or training-agnostic. Besides, considering the rapid development of pre-trained models, large pre-trained model-based OOD detection is also regarded as an important category and discussed separately. Furthermore, we provide a discussion of the evaluation scenarios, a variety of applications, and several future research directions. We believe this survey with new taxonomy will benefit the proposal of new methods and the expansion of more practical scenarios. A curated list of related papers is provided in the Github repository: https://github.com/shuolucs/Awesome-Out-Of-Distribution-Detection.

Dong Yan, Jian Liang, Yanbo Wang et al.

Test-Time Reinforcement Learning (TTRL) enables Large Language Models (LLMs) to enhance reasoning capabilities on unlabeled test streams by deriving pseudo-rewards from majority voting consensus. However, existing TTRL methods rely exclusively on positive pseudo-labeling strategies. Such reliance becomes vulnerable under challenging scenarios where answer distributions are highly dispersed, resulting in weak consensus that inadvertently reinforces incorrect trajectories as supervision signals. In this paper, we propose SCRL (Selective-Complementary Reinforcement Learning), a robust test-time reinforcement learning framework that effectively mitigates label noise amplification. SCRL develops Selective Positive Pseudo-Labeling, which enforces strict consensus criteria to filter unreliable majorities. Complementarily, SCRL introduces Entropy-Gated Negative Pseudo-Labeling, the first negative supervision mechanism in TTRL, to reliably prune incorrect trajectories based on generation uncertainty. Extensive experiments on multiple reasoning benchmarks demonstrate that SCRL achieves substantial improvements over baselines, while maintaining robust generalization and training stability under constrained rollout budgets. Our code is available at https://github.com/Jasper-Yan/SCRL.

Shuo Lu, Haohan Wang, Wei Feng et al.

Advertising image generation has increasingly focused on online metrics like Click-Through Rate (CTR), yet existing approaches adopt a ``one-size-fits-all" strategy that optimizes for overall CTR while neglecting preference diversity among user groups. This leads to suboptimal performance for specific groups, limiting targeted marketing effectiveness. To bridge this gap, we present \textit{One Size, Many Fits} (OSMF), a unified framework that aligns diverse group-wise click preferences in large-scale advertising image generation. OSMF begins with product-aware adaptive grouping, which dynamically organizes users based on their attributes and product characteristics, representing each group with rich collective preference features. Building on these groups, preference-conditioned image generation employs a Group-aware Multimodal Large Language Model (G-MLLM) to generate tailored images for each group. The G-MLLM is pre-trained to simultaneously comprehend group features and generate advertising images. Subsequently, we fine-tune the G-MLLM using our proposed Group-DPO for group-wise preference alignment, which effectively enhances each group's CTR on the generated images. To further advance this field, we introduce the Grouped Advertising Image Preference Dataset (GAIP), the first large-scale public dataset of group-wise image preferences, including around 600K groups built from 40M users. Extensive experiments demonstrate that our framework achieves the state-of-the-art performance in both offline and online settings. Our code and datasets will be released at https://github.com/JD-GenX/OSMF.

Yuan Xu, Jiabing Yang, Xiaofeng Wang et al.

Imitation learning based visuomotor policies have achieved strong performance in robotic manipulation, yet they often remain sensitive to egocentric viewpoint shifts. Unlike third-person viewpoint changes that only move the camera, egocentric shifts simultaneously alter both the camera pose and the robot action coordinate frame, making it necessary to jointly transfer action trajectories and synthesize corresponding observations under novel egocentric viewpoints. To address this challenge, we present EgoDemoGen, a framework that generates paired observation--action demonstrations under novel egocentric viewpoints through two key components: 1{)} EgoTrajTransfer, which transfers robot trajectories to the novel egocentric coordinate frame through motion-skill segmentation, geometry-aware transformation, and inverse kinematics filtering; and 2{)} EgoViewTransfer, a conditional video generation model that fuses a novel-viewpoint reprojected scene video and a robot motion video rendered from the transferred trajectory to synthesize photorealistic observations, trained with a self-supervised double reprojection strategy without requiring multi-viewpoint data. Experiments in simulation and real-world settings show that EgoDemoGen consistently improves policy success rates under both standard and novel egocentric viewpoints, with absolute gains of +24.6\% and +16.9\% in simulation and +16.0\% and +23.0\% on the real robot. Moreover, EgoViewTransfer achieves superior video generation quality for novel egocentric observations.

Huacan Wang, Ziyi Ni, Shuo Zhang et al.

The ultimate goal of code agents is to solve complex tasks autonomously. Although large language models (LLMs) have made substantial progress in code generation, real-world tasks typically demand full-fledged code repositories rather than simple scripts. Building such repositories from scratch remains a major challenge. Fortunately, GitHub hosts a vast, evolving collection of open-source repositories, which developers frequently reuse as modular components for complex tasks. Yet, existing frameworks like OpenHands and SWE-Agent still struggle to effectively leverage these valuable resources. Relying solely on README files provides insufficient guidance, and deeper exploration reveals two core obstacles: overwhelming information and tangled dependencies of repositories, both constrained by the limited context windows of current LLMs. To tackle these issues, we propose RepoMaster, an autonomous agent framework designed to explore and reuse GitHub repositories for solving complex tasks. For efficient understanding, RepoMaster constructs function-call graphs, module-dependency graphs, and hierarchical code trees to identify essential components, providing only identified core elements to the LLMs rather than the entire repository. During autonomous execution, it progressively explores related components using our exploration tools and prunes information to optimize context usage. Evaluated on the adjusted MLE-bench, RepoMaster achieves a 110% relative boost in valid submissions over the strongest baseline OpenHands. On our newly released GitTaskBench, RepoMaster lifts the task-pass rate from 40.7% to 62.9% while reducing token usage by 95%. Our code and demonstration materials are publicly available at https://github.com/QuantaAlpha/RepoMaster.

Ziyi Ni, Huacan Wang, Shuo Zhang et al.

Beyond scratch coding, exploiting large-scale code repositories (e.g., GitHub) for practical tasks is vital in real-world software development, yet current benchmarks rarely evaluate code agents in such authentic, workflow-driven scenarios. To bridge this gap, we introduce GitTaskBench, a benchmark designed to systematically assess this capability via 54 realistic tasks across 7 modalities and 7 domains. Each task pairs a relevant repository with an automated, human-curated evaluation harness specifying practical success criteria. Beyond measuring execution and task success, we also propose the alpha-value metric to quantify the economic benefit of agent performance, which integrates task success rates, token cost, and average developer salaries. Experiments across three state-of-the-art agent frameworks with multiple advanced LLMs show that leveraging code repositories for complex task solving remains challenging: even the best-performing system, OpenHands+Claude 3.7, solves only 48.15% of tasks (recent progress has pushed the frontier further, with RepoMaster+Claude 3.5 achieving a new record of 62.96%). Error analysis attributes over half of failures to seemingly mundane yet critical steps like environment setup and dependency resolution, highlighting the need for more robust workflow management and increased timeout preparedness. By releasing GitTaskBench, we aim to drive progress and attention toward repository-aware code reasoning, execution, and deployment -- moving agents closer to solving complex, end-to-end real-world tasks. The benchmark and code are open-sourced at https://github.com/QuantaAlpha/GitTaskBench.

Shuo Lu, Yanyin Chen, Wei Feng et al.

Layout generation plays a crucial role in enhancing both user experience and design efficiency. However, current approaches suffer from task-specific generation capabilities and perceptually misaligned evaluation metrics, leading to limited applicability and ineffective measurement. In this paper, we propose \textit{Uni-Layout}, a novel framework that achieves unified generation, human-mimicking evaluation and alignment between the two. For universal generation, we incorporate various layout tasks into a single taxonomy and develop a unified generator that handles background or element contents constrained tasks via natural language prompts. To introduce human feedback for the effective evaluation of layouts, we build \textit{Layout-HF100k}, the first large-scale human feedback dataset with 100,000 expertly annotated layouts. Based on \textit{Layout-HF100k}, we introduce a human-mimicking evaluator that integrates visual and geometric information, employing a Chain-of-Thought mechanism to conduct qualitative assessments alongside a confidence estimation module to yield quantitative measurements. For better alignment between the generator and the evaluator, we integrate them into a cohesive system by adopting Dynamic-Margin Preference Optimization (DMPO), which dynamically adjusts margins based on preference strength to better align with human judgments. Extensive experiments show that \textit{Uni-Layout} significantly outperforms both task-specific and general-purpose methods. Our code is publicly available at https://github.com/JD-GenX/Uni-Layout.

Shuo Lu, Jianjie Cheng, Yinuo Xu et al.

Multimodal large language models (MLLMs) have achieved strong performance on perception-oriented tasks, yet their ability to perform mathematical spatial reasoning, defined as the capacity to parse and manipulate two- and three-dimensional relations, remains unclear. Humans easily solve textbook-style spatial reasoning problems with over 95\% accuracy, but we find that most leading MLLMs fail to reach even 60\% on the same tasks. This striking gap highlights spatial reasoning as a fundamental weakness of current models. To investigate this gap, we present MathSpatial, a unified framework for evaluating and improving spatial reasoning in MLLMs. MathSpatial includes three complementary components: (i) MathSpatial-Bench, a benchmark of 2K problems across three categories and eleven subtypes, designed to isolate reasoning difficulty from perceptual noise; (ii) MathSpatial-Corpus, a training dataset of 8K additional problems with verified solutions; and (iii) MathSpatial-SRT, which models reasoning as structured traces composed of three atomic operations--Correlate, Constrain, and Infer. Experiments show that fine-tuning Qwen2.5-VL-7B on MathSpatial achieves competitive accuracy while reducing tokens by 25\%. MathSpatial provides the first large-scale resource that disentangles perception from reasoning, enabling precise measurement and comprehensive understanding of mathematical spatial reasoning in MLLMs.

Huizai Yao, Sicheng Zhao, Pengteng Li et al.

Source-Free Object Detection (SFOD) aims to adapt a source-pretrained object detector to a target domain without access to source data. However, existing SFOD methods predominantly rely on internal knowledge from the source model, which limits their capacity to generalize across domains and often results in biased pseudo-labels, thereby hindering both transferability and discriminability. In contrast, Vision Foundation Models (VFMs), pretrained on massive and diverse data, exhibit strong perception capabilities and broad generalization, yet their potential remains largely untapped in the SFOD setting. In this paper, we propose a novel SFOD framework that leverages VFMs as external knowledge sources to jointly enhance feature alignment and label quality. Specifically, we design three VFM-based modules: (1) Patch-weighted Global Feature Alignment (PGFA) distills global features from VFMs using patch-similarity-based weighting to enhance global feature transferability; (2) Prototype-based Instance Feature Alignment (PIFA) performs instance-level contrastive learning guided by momentum-updated VFM prototypes; and (3) Dual-source Enhanced Pseudo-label Fusion (DEPF) fuses predictions from detection VFMs and teacher models via an entropy-aware strategy to yield more reliable supervision. Extensive experiments on six benchmarks demonstrate that our method achieves state-of-the-art SFOD performance, validating the effectiveness of integrating VFMs to simultaneously improve transferability and discriminability.

Yingsheng Wang, Shuo Lu, Jian Liang et al.

Out-of-distribution (OOD) detection helps models identify data outside the training categories, crucial for security applications. While feature-based post-hoc methods address this by evaluating data differences in the feature space without changing network parameters, they often require access to training data, which may not be suitable for some data privacy scenarios. This may not be suitable in scenarios where data privacy protection is a concern. In this paper, we propose a simple yet effective post-hoc method, termed Classifier-based Feature Reconstruction (ClaFR), from the perspective of subspace projection. It first performs an orthogonal decomposition of the classifier's weights to extract the class-known subspace, then maps the original data features into this subspace to obtain new data representations. Subsequently, the OOD score is determined by calculating the feature reconstruction error of the data within the subspace. Compared to existing OOD detection algorithms, our method does not require access to training data while achieving leading performance on multiple OOD benchmarks. Our code is released at https://github.com/Aie0923/ClaFR.

Yinuo Xu, Shuo Lu, Jianjie Cheng et al.

Deep Research agents tackle knowledge-intensive tasks through multi-round retrieval and decision-oriented generation. While reinforcement learning (RL) has been shown to improve performance in this paradigm, its contributions remain underexplored. To fully understand the role of RL, we conduct a systematic study along three decoupled dimensions: prompt template, reward function, and policy optimization. Our study reveals that: 1) the Fast Thinking template yields greater stability and better performance than the Slow Thinking template used in prior work; 2) the F1-based reward underperforms the EM due to training collapse driven by answer avoidance; this can be mitigated by incorporating action-level penalties, ultimately surpassing EM; 3) REINFORCE outperforms PPO while requiring fewer search actions, whereas GRPO shows the poorest stability among policy optimization methods. Building on these insights, we then introduce Search-R1++, a strong baseline that improves the performance of Search-R1 from 0.403 to 0.442 (Qwen2.5-7B) and 0.289 to 0.331 (Qwen2.5-3B). We hope that our findings can pave the way for more principled and reliable RL training strategies in Deep Research systems.

Jie Xiao, Meng Chen, Qingnan Ren et al.

Reinforcement learning (RL) is a critical stage in post-training large language models (LLMs), involving repeated interaction between rollout generation, reward evaluation, and centralized learning. Distributing rollout execution offers opportunities to leverage more cost-efficient inference resources, but introduces challenges in wide-area coordination and policy dissemination. We present ECHO-2, a distributed RL framework for post-training with remote inference workers and non-negligible dissemination latency. ECHO-2 combines centralized learning with distributed rollouts and treats bounded policy staleness as a user-controlled parameter, enabling rollout generation, dissemination, and training to overlap. We introduce an overlap-based capacity model that relates training time, dissemination latency, and rollout throughput, yielding a practical provisioning rule for sustaining learner utilization. To mitigate dissemination bottlenecks and lower cost, ECHO-2 employs peer-assisted pipelined broadcast and cost-aware activation of heterogeneous workers. Experiments on GRPO post-training of 4B and 8B models under real wide-area bandwidth regimes show that ECHO-2 significantly improves cost efficiency while preserving RL reward comparable to strong baselines.

Run Ling, Wenji Wang, Yuting Liu et al.

Personalized image generation is crucial for improving the user experience, as it renders reference images into preferred ones according to user visual preferences. Although effective, existing methods face two main issues. First, existing methods treat all items in the user historical sequence equally when extracting user preferences, overlooking the varying semantic similarities between historical items and the reference item. Disproportionately high weights for low-similarity items distort users' visual preferences for the reference item. Second, existing methods heavily rely on consistency between generated and reference images to optimize the generation, which leads to underfitting user preferences and hinders personalization. To address these issues, we propose Retrieval Augment Personalized Image GenerAtion guided by Recommendation (RAGAR). Our approach uses a retrieval mechanism to assign different weights to historical items according to their similarities to the reference item, thereby extracting more refined users' visual preferences for the reference item. Then we introduce a novel rank task based on the multi-modal ranking model to optimize the personalization of the generated images instead of forcing depend on consistency. Extensive experiments and human evaluations on three real-world datasets demonstrate that RAGAR achieves significant improvements in both personalization and semantic metrics compared to five baselines.

Shuo Lu, Yinuo Xu, Jianjie Cheng et al.

Deep Research agents predominantly optimize search policies to maximize retrieval probability. However, we identify a critical bottleneck: the retrieval-utilization gap, where models fail to use gold evidence even after it is retrieved, due to context blindness in noisy environments. To bridge this gap, we propose DeepResearch-Slice, a simple yet effective neuro-symbolic framework. Unlike implicit attention, our approach predicts precise span indices to perform a deterministic hard filter before reasoning. Extensive evaluations across six benchmarks show substantial robustness gains. Applying our method to frozen backbones yields a 73 percent relative improvement, from 19.1 percent to 33.0 percent, effectively mitigating noise without requiring parameter updates to the reasoning model. These results highlight the need for explicit grounding mechanisms in open-ended research.

Huizai Yao, Sicheng Zhao, Shuo Lu et al.

Source-Free Object Detection (SFOD) enables knowledge transfer from a source domain to an unsupervised target domain for object detection without access to source data. Most existing SFOD approaches are either confined to conventional object detection (OD) models like Faster R-CNN or designed as general solutions without tailored adaptations for novel OD architectures, especially Detection Transformer (DETR). In this paper, we introduce Feature Reweighting ANd Contrastive Learning NetworK (FRANCK), a novel SFOD framework specifically designed to perform query-centric feature enhancement for DETRs. FRANCK comprises four key components: (1) an Objectness Score-based Sample Reweighting (OSSR) module that computes attention-based objectness scores on multi-scale encoder feature maps, reweighting the detection loss to emphasize less-recognized regions; (2) a Contrastive Learning with Matching-based Memory Bank (CMMB) module that integrates multi-level features into memory banks, enhancing class-wise contrastive learning; (3) an Uncertainty-weighted Query-fused Feature Distillation (UQFD) module that improves feature distillation through prediction quality reweighting and query feature fusion; and (4) an improved self-training pipeline with a Dynamic Teacher Updating Interval (DTUI) that optimizes pseudo-label quality. By leveraging these components, FRANCK effectively adapts a source-pre-trained DETR model to a target domain with enhanced robustness and generalization. Extensive experiments on several widely used benchmarks demonstrate that our method achieves state-of-the-art performance, highlighting its effectiveness and compatibility with DETR-based SFOD models.

Yongcan Yu, Lingxiao He, Shuo Lu et al.

Recent advances in vision-language models (VLMs) reasoning have been largely attributed to the rise of reinforcement Learning (RL), which has shifted the community's focus away from the supervised fine-tuning (SFT) paradigm. Many studies suggest that introducing the SFT stage not only fails to improve reasoning ability but may also negatively impact model training. In this study, we revisit this RL-centric belief through a systematic and controlled comparison of SFT and RL on VLM Reasoning. Using identical data sources, we find that the relative effectiveness of SFT and RL is conditional and strongly influenced by model capacity, data scale, and data distribution. Contrary to common assumptions, our findings show that SFT plays a crucial role across several scenarios: (1) Effectiveness for weaker models. SFT more reliably elicits reasoning capabilities in smaller or weaker VLMs. (2) Data efficiency. SFT with only 2K achieves comparable or better reasoning performance to RL with 20K. (3) Cross-modal transferability. SFT demonstrates stronger generalization across modalities. Moreover, we identify a pervasive issue of deceptive rewards, where higher rewards fail to correlate with better reasoning accuracy in RL. These results challenge the prevailing "RL over SFT" narrative. They highlight that the role of SFT may have been underestimated and support a more balanced post-training pipeline in which SFT and RL function as complementary components.