Yibo Li

Zhenhao Zhu, Yue Liu, Yanpei Guo et al.

We present GuardReasoner-Omni, a reasoning-based guardrail model designed to moderate text, image, video, and audio data. First, we construct a comprehensive training corpus comprising 181k samples spanning these four modalities. Our training pipeline follows a two-stage paradigm to incentivize the model to deliberate before making decisions: (1) conducting SFT to cold-start the model with explicit reasoning capabilities and structural adherence; and (2) performing RL with a concise correctness reward to preserve accurate reasoning while suppressing redundant generation. We release a suite of models scaled at 3B and 7B parameters. Extensive experiments demonstrate that GuardReasoner-Omni achieves superior performance compared to existing state-of-the-art baselines across various guardrail benchmarks.

Shuo Ji, Yibo Li, Bryan Hooi

Despite recent progress, LLM agents still struggle with reasoning over long interaction histories. While current memory-augmented agents rely on a static retrieve-then-reason paradigm, this rigid pipeline design prevents them from dynamically adapting memory access to intermediate evidence discovered during inference. To bridge this gap, we propose MRAgent, a framework that combines an associative memory graph with an active reconstruction mechanism. We represent memory as a Cue-Tag-Content graph, where associative tags serve as semantic bridges connecting fine-grained cues to memory contents. Operating on this structure, our active reconstruction mechanism integrates LLM reasoning directly into memory access, allowing the agent to iteratively explore and prune retrieval paths based on accumulated evidence. This ensures that memory retrieval is dynamically adapted to the reasoning context while avoiding combinatorial explosion caused by unconstrained expansion. Experiments on the LoCoMo benchmark and LongMemEval benchmark demonstrate significant improvements over strong baselines (up to 23%), while substantially reducing token and runtime cost, highlighting the effectiveness of active and associative reconstruction for long-horizon memory reasoning.

Yibo Li, Zijie Lin, Ailin Deng et al.

While Large Language Model (LLM) agents excel at general tasks, they inherently struggle with continual adaptation due to the frozen weights after deployment. Conventional reinforcement learning (RL) offers a solution but incurs prohibitive computational costs and the risk of catastrophic forgetting. We introduce Just-In-Time Reinforcement Learning (JitRL), a training-free framework that enables test-time policy optimization without any gradient updates. JitRL maintains a dynamic, non-parametric memory of experiences and retrieves relevant trajectories to estimate action advantages on-the-fly. These estimates are then used to directly modulate the LLM's output logits. We theoretically prove that this additive update rule is the exact closed-form solution to the KL-constrained policy optimization objective. Extensive experiments on WebArena and Jericho demonstrate that JitRL establishes a new state-of-the-art among training-free methods. Crucially, JitRL outperforms the performance of computationally expensive fine-tuning methods (e.g., WebRL) while reducing monetary costs by over 30 times, offering a scalable path for continual learning agents. The code is available at https://github.com/liushiliushi/JitRL.

Qianyun Guo, Yibo Li, Yue Liu et al.

Large Language Models (LLMs) are increasingly serving as personal assistants, where users share complex and diverse preferences over extended interactions. However, assessing how well LLMs can follow these preferences in realistic, long-term situations remains underexplored. This work proposes RealPref, a benchmark for evaluating realistic preference-following in personalized user-LLM interactions. RealPref features 100 user profiles, 1300 personalized preferences, four types of preference expression (ranging from explicit to implicit), and long-horizon interaction histories. It includes three types of test questions (multiple-choice, true-or-false, and open-ended), with detailed rubrics for LLM-as-a-judge evaluation. Results indicate that LLM performance significantly drops as context length grows and preference expression becomes more implicit, and that generalizing user preference understanding to unseen scenarios poses further challenges. RealPref and these findings provide a foundation for future research to develop user-aware LLM assistants that better adapt to individual needs. The code is available at https://github.com/GG14127/RealPref.

Yongshuai Huang, Ning Lu, Dapeng Chen et al.

Table structure recognition aims to extract the logical and physical structure of unstructured table images into a machine-readable format. The latest end-to-end image-to-text approaches simultaneously predict the two structures by two decoders, where the prediction of the physical structure (the bounding boxes of the cells) is based on the representation of the logical structure. However, the previous methods struggle with imprecise bounding boxes as the logical representation lacks local visual information. To address this issue, we propose an end-to-end sequential modeling framework for table structure recognition called VAST. It contains a novel coordinate sequence decoder triggered by the representation of the non-empty cell from the logical structure decoder. In the coordinate sequence decoder, we model the bounding box coordinates as a language sequence, where the left, top, right and bottom coordinates are decoded sequentially to leverage the inter-coordinate dependency. Furthermore, we propose an auxiliary visual-alignment loss to enforce the logical representation of the non-empty cells to contain more local visual details, which helps produce better cell bounding boxes. Extensive experiments demonstrate that our proposed method can achieve state-of-the-art results in both logical and physical structure recognition. The ablation study also validates that the proposed coordinate sequence decoder and the visual-alignment loss are the keys to the success of our method.

Yufei He, Juncheng Liu, Zhiyuan Hu et al.

Prevailing medical AI operates on an unrealistic ''one-shot'' model, diagnosing from a complete patient file. However, real-world diagnosis is an iterative inquiry where Clinicians sequentially ask questions and order tests to strategically gather information while managing cost and time. To address this, we first propose Med-Inquire, a new benchmark designed to evaluate an agent's ability to perform multi-turn diagnosis. Built upon a dataset of real-world clinical cases, Med-Inquire simulates the diagnostic process by hiding a complete patient file behind specialized Patient and Examination agents. They force the agent to proactively ask questions and order tests to gather information piece by piece. To tackle the challenges posed by Med-Inquire, we then introduce EvoClinician, a self-evolving agent that learns efficient diagnostic strategies at test time. Its core is a ''Diagnose-Grade-Evolve'' loop: an Actor agent attempts a diagnosis; a Process Grader agent performs credit assignment by evaluating each action for both clinical yield and resource efficiency; finally, an Evolver agent uses this feedback to update the Actor's strategy by evolving its prompt and memory. Our experiments show EvoClinician outperforms continual learning baselines and other self-evolving agents like memory agents. The code is available at https://github.com/yf-he/EvoClinician

Jiawei Liu, Cheng Yang, Zhiyuan Lu et al.

Foundation models have emerged as critical components in a variety of artificial intelligence applications, and showcase significant success in natural language processing and several other domains. Meanwhile, the field of graph machine learning is witnessing a paradigm transition from shallow methods to more sophisticated deep learning approaches. The capabilities of foundation models in generalization and adaptation motivate graph machine learning researchers to discuss the potential of developing a new graph learning paradigm. This paradigm envisions models that are pre-trained on extensive graph data and can be adapted for various graph tasks. Despite this burgeoning interest, there is a noticeable lack of clear definitions and systematic analyses pertaining to this new domain. To this end, this article introduces the concept of Graph Foundation Models (GFMs), and offers an exhaustive explanation of their key characteristics and underlying technologies. We proceed to classify the existing work related to GFMs into three distinct categories, based on their dependence on graph neural networks and large language models. In addition to providing a thorough review of the current state of GFMs, this article also outlooks potential avenues for future research in this rapidly evolving domain.

Yulin Chen, Tri Cao, Haoran Li et al.

Web agents powered by vision-language models (VLMs) enable autonomous interaction with web environments by perceiving and acting on both visual and textual webpage content to accomplish user-specified tasks. However, they are highly vulnerable to prompt injection attacks, where adversarial instructions embedded in HTML or rendered screenshots can manipulate agent behavior and lead to harmful outcomes such as information leakage. Existing defenses, including system prompt defenses and direct fine-tuning of agents, have shown limited effectiveness. To address this issue, we propose a defense framework in which a web agent operates in parallel with a dedicated guard agent, decoupling prompt injection detection from the agent's own reasoning. Building on this framework, we introduce WebAgentGuard, a reasoning-driven, multimodal guard model for prompt injection detection. We construct a synthetic multimodal dataset using GPT-5 spanning 164 topics and 230 visual and UI design styles, and train the model via reasoning-intensive supervised fine-tuning followed by reinforcement learning. Experiments across multiple benchmarks show that WebAgentGuard consistently outperforms strong baselines while preserving agent utility, without introducing additional latency.

Fangda Ye, Yuxin Hu, Pengxiang Zhu et al.

Recent progress in deep research systems has been impressive, but evaluation still lags behind real user needs. Existing benchmarks predominantly assess final reports using fixed rubrics, failing to evaluate the underlying research process. Most also offer limited multimodal coverage, rely on synthetic tasks that do not reflect real-world query complexity, and cannot be refreshed as knowledge evolves. To address these gaps, we introduce MiroEval, a benchmark and evaluation framework for deep research systems. The benchmark comprises 100 tasks (70 text-only, 30 multimodal), all grounded in real user needs and constructed via a dual-path pipeline that supports periodic updates, enabling a live and evolving setting. The proposed evaluation suite assesses deep research systems along three complementary dimensions: adaptive synthesis quality evaluation with task-specific rubrics, agentic factuality verification via active retrieval and reasoning over both web sources and multimodal attachments, and process-centric evaluation audits how the system searches, reasons, and refines throughout its investigation. Evaluation across 13 systems yields three principal findings: the three evaluation dimensions capture complementary aspects of system capability, with each revealing distinct strengths and weaknesses across systems; process quality serves as a reliable predictor of overall outcome while revealing weaknesses invisible to output-level metrics; and multimodal tasks pose substantially greater challenges, with most systems declining by 3 to 10 points. The MiroThinker series achieves the most balanced performance, with MiroThinker-H1 ranking the highest overall in both settings. Human verification and robustness results confirm the reliability of the benchmark and evaluation framework. MiroEval provides a holistic diagnostic tool for the next generation of deep research agents.

Yibo Li, Jiashuo Yang, Zhi Zheng et al.

LLM agents have shown strong performance across a wide range of complex tasks, including interactive environments that require long-horizon decision making. But these agents cannot learn on the fly at test time. Self-evolving agents address this by accumulating memory and reflection across episodes rather than requiring model-weight updates. However, these agents often suffer from exploration collapse: as memory grows, behavior concentrates around familiar high-reward routines, reducing the chance of discovering better alternatives. To address this problem, we propose Autonomous Policy EXploration (APEX), which builds and maintains an explicit strategy space through a strategy map-a directed acyclic graph of milestones with prerequisite dependency edges. In APEX, Fork Discovery expands the map with evidence-grounded unexplored directions, while Policy Selection balances exploration and exploitation during planning. Evaluated on nine Jericho text-adventure games and WebArena, a realistic web interaction benchmark, APEX outperforms all baselines. Extensive ablations validate each component's contribution and demonstrate robustness across diverse settings, demonstrating APEX's effectiveness for sustained exploration in self-evolving agents.

Yibo Li, Yuan Fang, Mengmei Zhang et al.

Understanding molecular structure and related knowledge is crucial for scientific research. Recent studies integrate molecular graphs with their textual descriptions to enhance molecular representation learning. However, they focus on the whole molecular graph and neglect frequently occurring subgraphs, known as motifs, which are essential for determining molecular properties. Without such fine-grained knowledge, these models struggle to generalize to unseen molecules and tasks that require motif-level insights. To bridge this gap, we propose FineMolTex, a novel Fine-grained Molecular graph-Text pre-training framework to jointly learn coarse-grained molecule-level knowledge and fine-grained motif-level knowledge. Specifically, FineMolTex consists of two pre-training tasks: a contrastive alignment task for coarse-grained matching and a masked multi-modal modeling task for fine-grained matching. In particular, the latter predicts the labels of masked motifs and words, which are selected based on their importance. By leveraging insights from both modalities, FineMolTex is able to understand the fine-grained matching between motifs and words. Finally, we conduct extensive experiments across three downstream tasks, achieving up to 230% improvement in the text-based molecule editing task. Additionally, our case studies reveal that FineMolTex successfully captures fine-grained knowledge, potentially offering valuable insights for drug discovery and catalyst design.

Yibo Li, Miao Xiong, Jiaying Wu et al.

Large Language Models (LLMs) are increasingly deployed in high-stakes domains such as science, law, and healthcare, where accurate expressions of uncertainty are essential for reliability and trust. However, current LLMs are often observed to generate incorrect answers with high confidence, a phenomenon known as "overconfidence". Recent efforts have focused on calibrating LLMs' verbalized confidence: i.e., their expressions of confidence in text form, such as "I am 80% confident that...". Existing approaches either rely on prompt engineering or fine-tuning with heuristically generated uncertainty estimates, both of which have limited effectiveness and generalizability. Motivated by the notion of proper scoring rules for calibration in classical machine learning models, we introduce ConfTuner, a simple and efficient fine-tuning method that introduces minimal overhead and does not require ground-truth confidence scores or proxy confidence estimates. ConfTuner relies on a new loss function, tokenized Brier score, which we theoretically prove to be a proper scoring rule, intuitively meaning that it "correctly incentivizes the model to report its true probability of being correct". ConfTuner improves calibration across diverse reasoning tasks and generalizes to black-box models such as GPT-4o. Our results further show that better-calibrated confidence enables downstream gains in self-correction and model cascade, advancing the development of trustworthy LLM systems. The code is available at https://github.com/liushiliushi/ConfTuner.

Tri Cao, Yulin Chen, Hieu Cao et al.

Web agents can autonomously complete online tasks by interacting with websites, but their exposure to open web environments makes them vulnerable to prompt injection attacks embedded in HTML content or visual interfaces. Existing guard models still suffer from limited generalization to unseen domains and attack patterns, high false positive rates on benign content, reduced deployment efficiency due to added latency at each step, and vulnerability to adversarial attacks that evolve over time or directly target the guard itself. To address these limitations, we propose WARD (Web Agent Robust Defense against Prompt Injection), a practical guard model for secure and efficient web agents. WARD is built on WARD-Base, a large-scale dataset with around 177K samples collected from 719 high-traffic URLs and platforms, and WARD-PIG, a dedicated dataset designed for prompt injection attacks targeting the guard model. We further introduce A3T, an adaptive adversarial attack training framework that iteratively strengthens WARD through a memory-based attacker and guard co-evolution process. Extensive experiments show that WARD achieves nearly perfect recall on out-of-distribution benchmarks, maintains low false positive rates to preserve agent utility, remains robust against guard-targeted and adaptive attacks under substantial distribution shifts, and runs efficiently in parallel with the agent without introducing additional latency.

Huiqi Deng, Yibo Li, Quanshi Zhang et al.

Understanding why deep neural networks (DNNs) fail to generalize to unseen samples remains a long-standing challenge. Existing studies mainly examine changes in externally observable factors such as data, representations, or outputs, yet offer limited insight into how a model's internal decision mechanism evolves from training to test. To address this gap, we introduce Decision Pattern Shift (DPS), a new perspective that defines generalization through the stability of internal decision patterns and quantifies failure as their deviation from those learned during training. Specifically, we represent each sample's decision pattern as a GradCAM-based channel-contribution vector, which captures how feature channels collectively support a prediction, and we propose the DPS metric to measure its discrepancy from the class-average pattern. Empirical analyses across multiple datasets and architectures show that, (i) decision patterns form a highly structured, class-consistent space with strong intra-class cohesion and low inter-class confusion, enabling direct analysis of a model's decision logic; (ii) the DPS magnitude correlates linearly with the generalization gap (nearly all Pearson r > 0.8), revealing generalization as a systematic drift in the model's internal decision mechanism; (iii) the DPS spectrum organizes diverse generalization degradation scenarios (covering ideal generalization, in-distribution degradation, domain shift, out-of-distribution, and shortcut learning) into a continuous trajectory, providing a unified explanation of their failure modes. These findings open up new possibilities for early generalization-risk detection, failure-mode diagnosis, and channel-level defect localization.

Yanxing Wang, Jianxing Hu, Junyong Lai et al.

Molecular fingerprints are the workhorse in ligand-based drug discovery. In recent years, an increasing number of research papers reported fascinating results on using deep neural networks to learn 2D molecular representations as fingerprints. It is anticipated that the integration of deep learning would also contribute to the prosperity of 3D fingerprints. Here, we unprecedentedly introduce deep learning into 3D small molecule fingerprints, presenting a new one we termed as the three-dimensional force fields fingerprint (TF3P). TF3P is learned by a deep capsular network whose training is in no need of labeled datasets for specific predictive tasks. TF3P can encode the 3D force fields information of molecules and demonstrates the stronger ability to capture 3D structural changes, to recognize molecules alike in 3D but not in 2D, and to identify similar targets inaccessible by other 2D or 3D fingerprints based on only ligands similarity. Furthermore, TF3P is compatible with both statistical models (e.g. similarity ensemble approach) and machine learning models. Altogether, we report TF3P as a new 3D small molecule fingerprint with a promising future in ligand-based drug discovery. All codes are written in Python and available at https://github.com/canisw/tf3p.

Yibo Li, Qiongxiu Li

Gradient inversion attacks reveal that private training text can be reconstructed from shared gradients, posing a privacy risk to large language models (LLMs). While prior methods perform well in small-batch settings, scaling to larger batch sizes and longer sequences remains challenging due to severe signal mixing, high computational cost, and degraded fidelity. We present SOMP (Subspace-Guided Orthogonal Matching Pursuit), a scalable gradient inversion framework that casts text recovery from aggregated gradients as a sparse signal recovery problem. Our key insight is that aggregated transformer gradients retain exploitable head-wise geometric structure together with sample-level sparsity. SOMP leverages these properties to progressively narrow the search space and disentangle mixed signals without exhaustive search. Experiments across multiple LLM families, model scales, and five languages show that SOMP consistently outperforms prior methods in the aggregated-gradient regime.For long sequences at batch size B=16, SOMP achieves substantially higher reconstruction fidelity than strong baselines, while remaining computationally competitive. Even under extreme aggregation (up to B=128), SOMP still recovers meaningful text, suggesting that privacy leakage can persist in regimes where prior attacks become much less effective.

Rui Yan, Yibo Li, Han Ding et al.

Electroencephalogram (EEG)-based emotion recognition is vital for affective computing but faces challenges in feature utilization and cross-domain generalization. This work introduces EmotionCLIP, which reformulates recognition as an EEG-text matching task within the CLIP framework. A tailored backbone, SST-LegoViT, captures spatial, spectral, and temporal features using multi-scale convolution and Transformer modules. Experiments on SEED and SEED-IV datasets show superior cross-subject accuracies of 88.69% and 73.50%, and cross-time accuracies of 88.46% and 77.54%, outperforming existing models. Results demonstrate the effectiveness of multimodal contrastive learning for robust EEG emotion recognition.

Yuan Sui, Yulin Chen, Yibo Li et al.

When language model agents tackle complex software engineering tasks, they often degrade over long trajectories, which we define as *agent drift*. We focus on two recurring failure modes *overthinking* and *overacting*, i.e., where the agent repeatedly reasons over information it already has, and where it issues tool calls without integrating recent observations or acquiring new evidence. In this paper, we introduce TACT (Think-Act Calibration via activation Steering), to detect and mitigate agent drift in the residual stream before it surfaces as a behavioral failure. In specific, we label trajectory steps as overthinking, overacting, or calibrated, and find that their hidden states can separate linearly along two *drift axes*, pointing from calibrated behavior toward each failure mode (AUC $\approx$ 0.9). To mitigate agent drift, we project each step's activation onto these axes at test time and pull drifted ones back toward the calibrated region. Experiments show that TACT outperforms unsteered baselines across SWE-bench Verified, Terminal-Bench 2.0, and CLAW-Eval, lifting average resolve rate by $+5.8$ pp on Qwen3.5-27B and $+4.8$ pp on Gemma-4-26B-A4B-it while cutting steps-to-resolve by up to $26\%$. These gains frame agent drift as a steerable direction in the residual stream, and position TACT as a viable handle for reliable long-horizon agents.

Yibo Li, Xiao Wang, Hongrui Liu et al.

Recent studies reveal the connection between GNNs and the diffusion process, which motivates many diffusion-based GNNs to be proposed. However, since these two mechanisms are closely related, one fundamental question naturally arises: Is there a general diffusion framework that can formally unify these GNNs? The answer to this question can not only deepen our understanding of the learning process of GNNs, but also may open a new door to design a broad new class of GNNs. In this paper, we propose a general diffusion equation framework with the fidelity term, which formally establishes the relationship between the diffusion process with more GNNs. Meanwhile, with this framework, we identify one characteristic of graph diffusion networks, i.e., the current neural diffusion process only corresponds to the first-order diffusion equation. However, by an experimental investigation, we show that the labels of high-order neighbors actually exhibit monophily property, which induces the similarity based on labels among high-order neighbors without requiring the similarity among first-order neighbors. This discovery motives to design a new high-order neighbor-aware diffusion equation, and derive a new type of graph diffusion network (HiD-Net) based on the framework. With the high-order diffusion equation, HiD-Net is more robust against attacks and works on both homophily and heterophily graphs. We not only theoretically analyze the relation between HiD-Net with high-order random walk, but also provide a theoretical convergence guarantee. Extensive experimental results well demonstrate the effectiveness of HiD-Net over state-of-the-art graph diffusion networks.

Yibo Li, Xiao Wang, Yujie Xing et al.

Graph neural networks (GNNs) have achieved remarkable performance on graph-structured data. However, GNNs may inherit prejudice from the training data and make discriminatory predictions based on sensitive attributes, such as gender and race. Recently, there has been an increasing interest in ensuring fairness on GNNs, but all of them are under the assumption that the training and testing data are under the same distribution, i.e., training data and testing data are from the same graph. Will graph fairness performance decrease under distribution shifts? How does distribution shifts affect graph fairness learning? All these open questions are largely unexplored from a theoretical perspective. To answer these questions, we first theoretically identify the factors that determine bias on a graph. Subsequently, we explore the factors influencing fairness on testing graphs, with a noteworthy factor being the representation distances of certain groups between the training and testing graph. Motivated by our theoretical analysis, we propose our framework FatraGNN. Specifically, to guarantee fairness performance on unknown testing graphs, we propose a graph generator to produce numerous graphs with significant bias and under different distributions. Then we minimize the representation distances for each certain group between the training graph and generated graphs. This empowers our model to achieve high classification and fairness performance even on generated graphs with significant bias, thereby effectively handling unknown testing graphs. Experiments on real-world and semi-synthetic datasets demonstrate the effectiveness of our model in terms of both accuracy and fairness.

Yufei He, Juncheng Liu, Yue Liu et al.

A fundamental limitation of current AI agents is their inability to learn complex skills on the fly at test time, often behaving like "clever but clueless interns" in novel environments. This severely limits their practical utility. To systematically measure and drive progress on this challenge, we first introduce the Jericho Test-Time Learning (J-TTL) benchmark. J-TTL is a new evaluation setup where an agent must play the same game for several consecutive episodes, attempting to improve its performance from one episode to the next. On J-TTL, we find that existing adaptation methods like reflection, memory, or reinforcement learning struggle. To address the challenges posed by our benchmark, we present EvoTest, an evolutionary test-time learning framework that improves an agent without any fine-tuning or gradients-by evolving the entire agentic system after every episode. EvoTest has two roles: the Actor Agent, which plays the game, and the Evolver Agent, which analyzes the episode transcript to propose a revised configuration for the next run. This configuration rewrites the prompt, updates memory by logging effective state-action choices, tunes hyperparameters, and learns the tool-use routines. On our J-TTL benchmark, EvoTest consistently increases performance, outperforming not only reflection and memory-only baselines but also more complex online fine-tuning methods. Notably, our method is the only one capable of winning two games (Detective and Library), while all baselines fail to win any.

Zhanzhi Lou, Hui Chen, Yibo Li et al.

Test-Time Learning (TTL) enables language agents to iteratively refine their performance through repeated interactions with the environment at inference time. At the core of TTL is an adaptation policy that updates the actor policy based on experience from previous episodes, thereby improving future behavior. Existing methods rely on fixed, hand-crafted adaptation policies rather than optimizing them for downstream improvement. We argue that optimal adaptation policies should be learned from task environments, not hand-engineered based on human intuition. To achieve this, we introduce Meta-TTL, a framework that formulates the discovery of effective adaptation policies as a bi-level optimization problem. Within this framework, the inner loop executes the standard TTL process, measuring how effectively a candidate adaptation policy helps an agent correct errors across sequential episodes. Guided by the agent's performance, the outer loop employs evolutionary search over a diverse distribution of training tasks to iteratively refine the adaptation policy. We evaluate Meta-TTL on Jericho and WebArena-Lite across both in-distribution (ID) and out-of-distribution (OOD) settings, using multiple meta-agent backbones. Results on both benchmarks show that Meta-TTL consistently outperforms hand-crafted baselines, suggesting that the optimized adaptation policy encodes transferable strategies that generalize beyond the training task distribution.

Enshen Zhou, Cheng Chi, Yibo Li et al.

Spatial tracing, as a fundamental embodied interaction ability for robots, is inherently challenging as it requires multi-step metric-grounded reasoning compounded with complex spatial referring and real-world metric measurement. However, existing methods struggle with this compositional task. To this end, we propose RoboTracer, a 3D-aware VLM that first achieves both 3D spatial referring and measuring via a universal spatial encoder and a regression-supervised decoder to enhance scale awareness during supervised fine-tuning (SFT). Moreover, RoboTracer advances multi-step metric-grounded reasoning via reinforcement fine-tuning (RFT) with metric-sensitive process rewards, supervising key intermediate perceptual cues to accurately generate spatial traces. To support SFT and RFT training, we introduce TraceSpatial, a large-scale dataset of 30M QA pairs, spanning outdoor/indoor/tabletop scenes and supporting complex reasoning processes (up to 9 steps). We further present TraceSpatial-Bench, a challenging benchmark filling the gap to evaluate spatial tracing. Experimental results show that RoboTracer surpasses baselines in spatial understanding, measuring, and referring, with an average success rate of 79.1%, and also achieves SOTA performance on TraceSpatial-Bench by a large margin, exceeding Gemini-2.5-Pro by 36% accuracy. Notably, RoboTracer can be integrated with various control policies to execute long-horizon, dynamic tasks across diverse robots (UR5, G1 humanoid) in cluttered real-world scenes. See the project page at https://zhoues.github.io/RoboTracer.

Yi Liu, Xiao Xu, Zeyu Xu et al.

Vision-Language Models (VLMs) have achieved remarkable breakthroughs in recent years, enabling a diverse array of applications in everyday life. However, the substantial computational and storage demands of VLMs pose significant challenges for their efficient deployment on mobile devices, which represent the most ubiquitous and accessible computing platforms today. In this work, we introduce MagicVL-2B, a novel VLM meticulously optimized for flagship smartphones. MagicVL-2B leverages a lightweight visual encoder with fewer than 100M parameters and features a redesigned dynamic resolution scheme that adaptively generates image tokens without excessive modification of image dimensions. To further enhance the performance of this compact encoder within VLMs, we propose a multimodal curriculum learning strategy that incrementally increases task difficulty and data information density throughout training. This approach substantially improves the model's performance across a variety of sub-tasks. Extensive evaluations on standard VLM benchmarks demonstrate that MagicVL-2B matches the accuracy of current state-of-the-art models while reducing on-device power consumption by 41.1%. These results establish MagicVL-2B as a practical and robust solution for real-world mobile vision-language applications, enabling advanced multimodal intelligence to run directly on smartphones.

Yujie Xing, Xiao Wang, Yibo Li et al.

Graph Transformer, due to its global attention mechanism, has emerged as a new tool in dealing with graph-structured data. It is well recognized that the global attention mechanism considers a wider receptive field in a fully connected graph, leading many to believe that useful information can be extracted from all the nodes. In this paper, we challenge this belief: does the globalizing property always benefit Graph Transformers? We reveal the over-globalizing problem in Graph Transformer by presenting both empirical evidence and theoretical analysis, i.e., the current attention mechanism overly focuses on those distant nodes, while the near nodes, which actually contain most of the useful information, are relatively weakened. Then we propose a novel Bi-Level Global Graph Transformer with Collaborative Training (CoBFormer), including the inter-cluster and intra-cluster Transformers, to prevent the over-globalizing problem while keeping the ability to extract valuable information from distant nodes. Moreover, the collaborative training is proposed to improve the model's generalization ability with a theoretical guarantee. Extensive experiments on various graphs well validate the effectiveness of our proposed CoBFormer.

Tianyu Zhao, Cheng Yang, Yibo Li et al.

Heterogeneous Graph Neural Network (HGNN) has been successfully employed in various tasks, but we cannot accurately know the importance of different design dimensions of HGNNs due to diverse architectures and applied scenarios. Besides, in the research community of HGNNs, implementing and evaluating various tasks still need much human effort. To mitigate these issues, we first propose a unified framework covering most HGNNs, consisting of three components: heterogeneous linear transformation, heterogeneous graph transformation, and heterogeneous message passing layer. Then we build a platform Space4HGNN by defining a design space for HGNNs based on the unified framework, which offers modularized components, reproducible implementations, and standardized evaluation for HGNNs. Finally, we conduct experiments to analyze the effect of different designs. With the insights found, we distill a condensed design space and verify its effectiveness.

Ke Yuan, Zuoyu Yan, Yibo Li et al.

Math expressions are important parts of scientific and educational documents, but some of them may be challenging for junior scholars or students to understand. Nevertheless, constructing textual descriptions for math expressions is nontrivial. In this paper, we explore the feasibility to automatically construct descriptions for math expressions. But there are two challenges that need to be addressed: 1) finding relevant documents since a math equation understanding usually requires several topics, but these topics are often explained in different documents. 2) the sparsity of the collected relevant documents making it difficult to extract reasonable descriptions. Different documents mainly focus on different topics which makes model hard to extract salient information and organize them to form a description of math expressions. To address these issues, we propose a hybrid model (MathDes) which contains two important modules: Selector and Summarizer. In the Selector, a Topic Relation Graph (TRG) is proposed to obtain the relevant documents which contain the comprehensive information of math expressions. TRG is a graph built according to the citations between expressions. In the Summarizer, a summarization model under the Integer Linear Programming (ILP) framework is proposed. This module constructs the final description with the help of a timeline that is extracted from TRG. The experimental results demonstrate that our methods are promising for this task and outperform the baselines in all aspects.

Yibo Li, Jianfeng Pei, Luhua Lai

Recently, deep generative models for molecular graphs are gaining more and more attention in the field of de novo drug design. A variety of models have been developed to generate topological structures of drug-like molecules, but explorations in generating three-dimensional structures are still limited. Existing methods have either focused on low molecular weight compounds without considering drug-likeness or generate 3D structures indirectly using atom density maps. In this work, we introduce Ligand Neural Network (L-Net), a novel graph generative model for designing drug-like molecules with high-quality 3D structures. L-Net directly outputs the topological and 3D structure of molecules (including hydrogen atoms), without the need for additional atom placement or bond order inference algorithm. The architecture of L-Net is specifically optimized for drug-like molecules, and a set of metrics is assembled to comprehensively evaluate its performance. The results show that L-Net is capable of generating chemically correct, conformationally valid, and highly druglike molecules. Finally, to demonstrate its potential in structure-based molecular design, we combine L-Net with MCTS and test its ability to generate potential inhibitors targeting ABL1 kinase.

Yibo Li, Jianxing Hu, Yanxing Wang et al.

The ultimate goal of drug design is to find novel compounds with desirable pharmacological properties. Designing molecules retaining particular scaffolds as the core structures of the molecules is one of the efficient ways to obtain potential drug candidates with desirable properties. We proposed a scaffold-based molecular generative model for scaffold-based drug discovery, which performs molecule generation based on a wide spectrum of scaffold definitions, including BM-scaffolds, cyclic skeletons, as well as scaffolds with specifications on side-chain properties. The model can generalize the learned chemical rules of adding atoms and bonds to a given scaffold. Furthermore, the generated compounds were evaluated by molecular docking in DRD2 targets and the results demonstrated that this approach can be effectively applied to solve several drug design problems, including the generation of compounds containing a given scaffold and de novo drug design of potential drug candidates with specific docking scores. Finally, a command line interface is created.

Yibo Li, Liangren Zhang, Zhenming Liu

Recently, deep generative models have revealed itself as a promising way of performing de novo molecule design. However, previous research has focused mainly on generating SMILES strings instead of molecular graphs. Although current graph generative models are available, they are often too general and computationally expensive, which restricts their application to molecules with small sizes. In this work, a new de novo molecular design framework is proposed based on a type sequential graph generators that do not use atom level recurrent units. Compared with previous graph generative models, the proposed method is much more tuned for molecule generation and have been scaled up to cover significantly larger molecules in the ChEMBL database. It is shown that the graph-based model outperforms SMILES based models in a variety of metrics, especially in the rate of valid outputs. For the application of drug design tasks, conditional graph generative model is employed. This method offers higher flexibility compared to previous fine-tuning based approach and is suitable for generation based on multiple objectives. This approach is applied to solve several drug design problems, including the generation of compounds containing a given scaffold, generation of compounds with specific drug-likeness and synthetic accessibility requirements, as well as generating dual inhibitors against JNK3 and GSK3$β$. Results show high enrichment rates for outputs satisfying the given requirements.