Ziyuan Wang

GLM-5 Team, Aohan Zeng, Xin Lv et al. · tsinghua

We present GLM-5, a next-generation foundation model designed to transition the paradigm of vibe coding to agentic engineering. Building upon the agentic, reasoning, and coding (ARC) capabilities of its predecessor, GLM-5 adopts DSA to significantly reduce training and inference costs while maintaining long-context fidelity. To advance model alignment and autonomy, we implement a new asynchronous reinforcement learning infrastructure that drastically improves post-training efficiency by decoupling generation from training. Furthermore, we propose novel asynchronous agent RL algorithms that further improve RL quality, enabling the model to learn from complex, long-horizon interactions more effectively. Through these innovations, GLM-5 achieves state-of-the-art performance on major open benchmarks. Most critically, GLM-5 demonstrates unprecedented capability in real-world coding tasks, surpassing previous baselines in handling end-to-end software engineering challenges. Code, models, and more information are available at https://github.com/zai-org/GLM-5.

Biyang Guo, Xin Zhang, Ziyuan Wang et al.

The introduction of ChatGPT has garnered widespread attention in both academic and industrial communities. ChatGPT is able to respond effectively to a wide range of human questions, providing fluent and comprehensive answers that significantly surpass previous public chatbots in terms of security and usefulness. On one hand, people are curious about how ChatGPT is able to achieve such strength and how far it is from human experts. On the other hand, people are starting to worry about the potential negative impacts that large language models (LLMs) like ChatGPT could have on society, such as fake news, plagiarism, and social security issues. In this work, we collected tens of thousands of comparison responses from both human experts and ChatGPT, with questions ranging from open-domain, financial, medical, legal, and psychological areas. We call the collected dataset the Human ChatGPT Comparison Corpus (HC3). Based on the HC3 dataset, we study the characteristics of ChatGPT's responses, the differences and gaps from human experts, and future directions for LLMs. We conducted comprehensive human evaluations and linguistic analyses of ChatGPT-generated content compared with that of humans, where many interesting results are revealed. After that, we conduct extensive experiments on how to effectively detect whether a certain text is generated by ChatGPT or humans. We build three different detection systems, explore several key factors that influence their effectiveness, and evaluate them in different scenarios. The dataset, code, and models are all publicly available at https://github.com/Hello-SimpleAI/chatgpt-comparison-detection.

Fangyuan Wang, Ziyuan Wang, Guorui Pei et al.

Robotic manipulation depends on the ability to anticipate how actions reshape objects, contacts, and scene geometry before execution. Learned world models provide this capability by predicting task-relevant future evolution under robot intervention, yet the term now spans latent dynamics models, action-conditioned video generators, three- and four-dimensional scene predictors, physics-informed simulators, and predictive modules inside vision-language-action systems. This breadth has fragmented the literature and obscured the design choices that matter for manipulation. We survey world models for robotic manipulation through three questions: what future representation is predicted, how prediction is connected to action, and when prediction is used in the robot-learning pipeline. We operationally define a world model as an action-conditioned predictive system and distinguish it from perception modules, inverse models, policies, rewards, and value functions. We then organize existing work into five representation families, develop a functional taxonomy that separates integrated prediction-action models from explicit predictive planners, and characterize infrastructure roles including synthetic experience generation, candidate filtering, search-based evaluation, learned environments, and outcome verification. We further map these roles across pretraining, post-training, and inference adaptation, review 34 manipulation datasets, and synthesize evaluation protocols for predictive fidelity, task performance, and simulator reliability. This survey shows that world models are evolving from task-specific dynamics predictors into predictive infrastructure for robot learning, while exposing open challenges in contact modeling, hallucination control, action alignment, and benchmarking under closed-loop use.

Ziyuan Wang, Hailiang Huang, Songqiao Han

Current text classification methods typically encode the text merely into embedding before a naive or complicated classifier, which ignores the suggestive information contained in the label text. As a matter of fact, humans classify documents primarily based on the semantic meaning of the subcategories. We propose a novel model structure via siamese BERT and interactive double attentions named IDEA ( Interactive DoublE Attentions) to capture the information exchange of text and label names. Interactive double attentions enable the model to exploit the inter-class and intra-class information from coarse to fine, which involves distinguishing among all labels and matching the semantical subclasses of ground truth labels. Our proposed method outperforms the state-of-the-art methods using label texts significantly with more stable results.

Thomas Langerak, Renate Zhang, Ziyuan Wang et al.

Deciding which idea is worth prototyping is a central concern in iterative design. A prototype should be produced when the expected improvement is high and the cost is low. However, this is hard to decide, because costs can vary drastically: a simple parameter tweak may take seconds, while fabricating hardware consumes material and energy. Such asymmetries, can discourage a designer from exploring the design space. In this paper, we present an extension of cost-aware Bayesian optimization to account for diverse prototyping costs. The method builds on the power of Bayesian optimization and requires only a minimal modification to the acquisition function. The key idea is to use designer-estimated costs to guide sampling toward more cost-effective prototypes. In technical evaluations, the method achieved comparable utility to a cost-agnostic baseline while requiring only ${\approx}70\%$ of the cost; under strict budgets, it outperformed the baseline threefold. A within-subjects study with 12 participants in a realistic joystick design task demonstrated similar benefits. These results show that accounting for prototyping costs can make Bayesian optimization more compatible with real-world design projects.

Hanbing Liu, Chunhao Tian, Nan An et al.

We study budget-constrained tool-augmented agents, where a large language model must solve multi-step tasks by invoking external tools under a strict monetary budget. We formalize this setting as sequential decision making in context space with priced and stochastic tool executions, making direct planning intractable due to massive state-action spaces, high variance of outcomes and prohibitive exploration cost. To address these challenges, we propose INTENT, an inference-time planning framework that leverages an intention-aware hierarchical world model to anticipate future tool usage, risk-calibrated cost, and guide decisions online. Across cost-augmented StableToolBench, INTENT strictly enforces hard budget feasibility while substantially improving task success over baselines, and remains robust under dynamic market shifts such as tool price changes and varying budgets.

Vanshika Vats, Marzia Binta Nizam, Minghao Liu et al.

As the capabilities of artificial intelligence (AI) continue to expand rapidly, Human-AI (HAI) Collaboration, combining human intellect and AI systems, has become pivotal for advancing problem-solving and decision-making processes. The advent of Large Foundation Models (LFMs) has greatly expanded its potential, offering unprecedented capabilities by leveraging vast amounts of data to understand and predict complex patterns. At the same time, realizing this potential responsibly requires addressing persistent challenges related to safety, fairness, and control. This paper reviews the crucial integration of LFMs with HAI, highlighting both opportunities and risks. We structure our analysis around four areas: human-guided model development, collaborative design principles, ethical and governance frameworks, and applications in high-stakes domains. Our review shows that successful HAI systems are not the automatic result of stronger models but the product of careful, human-centered design. By identifying key open challenges, this survey aims to give insight into current and future research that turns the raw power of LFMs into partnerships that are reliable, trustworthy, and beneficial to society.

Zitao Wang, Ziyuan Wang, Molei Liu et al.

Transfer learning is a popular strategy to leverage external knowledge and improve statistical efficiency, particularly with a limited target sample. We propose a novel knowledge-guided Wasserstein Distributionally Robust Optimization (KG-WDRO) framework that adaptively incorporates multiple sources of external knowledge to overcome the conservativeness of vanilla WDRO, which often results in overly pessimistic shrinkage toward zero. Our method constructs smaller Wasserstein ambiguity sets by controlling the transportation along directions informed by the source knowledge. This strategy can alleviate perturbations on the predictive projection of the covariates and protect against information loss. Theoretically, we establish the equivalence between our WDRO formulation and the knowledge-guided shrinkage estimation based on collinear similarity, ensuring tractability and geometrizing the feasible set. This also reveals a novel and general interpretation for recent shrinkage-based transfer learning approaches from the perspective of distributional robustness. In addition, our framework can adjust for scaling differences in the regression models between the source and target and accommodates general types of regularization such as lasso and ridge. Extensive simulations demonstrate the superior performance and adaptivity of KG-WDRO in enhancing small-sample transfer learning.

Genliang Zhu, Chu Wang, Ziyuan Wang et al.

AI agents increasingly require direct, structured access to application data and actions, but production deployments still struggle to express and verify the governance properties that matter in practice: least-privilege authorization, controlled write execution, predictable failure handling, abuse resistance, and auditability. This paper introduces OpenPort Protocol (OPP), a governance-first specification for exposing application tools through a secure server-side gateway that is model- and runtime-neutral and can bind to existing tool ecosystems. OpenPort defines authorization-dependent discovery, stable response envelopes with machine-actionable \texttt{agent.*} reason codes, and an authorization model combining integration credentials, scoped permissions, and ABAC-style policy constraints. For write operations, OpenPort specifies a risk-gated lifecycle that defaults to draft creation and human review, supports time-bounded auto-execution under explicit policy, and enforces high-risk safeguards including preflight impact binding and idempotency. To address time-of-check/time-of-use drift in delayed approval flows, OpenPort also specifies an optional State Witness profile that revalidates execution-time preconditions and fails closed on state mismatch. Operationally, the protocol requires admission control (rate limits/quotas) with stable 429 semantics and structured audit events across allow/deny/fail paths so that client recovery and incident analysis are deterministic. We present a reference runtime and an executable governance toolchain (layered conformance profiles, negative security tests, fuzz/abuse regression, and release-gate scans) and evaluate the core profile at a pinned release tag using artifact-based, externally reproducible validation.

Kai Chen, Xiaodong Zhao, Yujie Huang et al.

The analysis and prediction of agent trajectories are crucial for decision-making processes in intelligent systems, with precise short-term trajectory forecasting being highly significant across a range of applications. Agents and their social interactions have been quantified and modeled by researchers from various perspectives; however, substantial limitations exist in the current work due to the inherent high uncertainty of agent intentions and the complex higher-order influences among neighboring groups. SocialMOIF is proposed to tackle these challenges, concentrating on the higher-order intention interactions among neighboring groups while reinforcing the primary role of first-order intention interactions between neighbors and the target agent. This method develops a multi-order intention fusion model to achieve a more comprehensive understanding of both direct and indirect intention information. Within SocialMOIF, a trajectory distribution approximator is designed to guide the trajectories toward values that align more closely with the actual data, thereby enhancing model interpretability. Furthermore, a global trajectory optimizer is introduced to enable more accurate and efficient parallel predictions. By incorporating a novel loss function that accounts for distance and direction during training, experimental results demonstrate that the model outperforms previous state-of-the-art baselines across multiple metrics in both dynamic and static datasets.

Yifan Yang, Zheshu Song, Jianheng Zhuo et al.

The evolution of speech technology has been spurred by the rapid increase in dataset sizes. Traditional speech models generally depend on a large amount of labeled training data, which is scarce for low-resource languages. This paper presents GigaSpeech 2, a large-scale, multi-domain, multilingual speech recognition corpus. It is designed for low-resource languages and does not rely on paired speech and text data. GigaSpeech 2 comprises about 30,000 hours of automatically transcribed speech, including Thai, Indonesian, and Vietnamese, gathered from unlabeled YouTube videos. We also introduce an automated pipeline for data crawling, transcription, and label refinement. Specifically, this pipeline involves Whisper for initial transcription, MMS for forced alignment, and multi-dimensional filtering for data quality assurance. A modified Noisy Student Training is developed to further refine flawed pseudo labels iteratively, thereby enhancing model performance. Experimental results on our manually transcribed evaluation set and two public test sets from Common Voice and FLEURS confirm our corpus's high quality and broad applicability. Notably, ASR models trained on GigaSpeech 2 can reduce the word error rate for Thai, Indonesian, and Vietnamese on our challenging and realistic YouTube test set by 25% to 40% compared to Whisper large-v3, with merely 10% model parameters. Furthermore, our ASR models trained on GigaSpeech 2 yield superior performance compared to commercial services. We hope that our newly introduced corpus and pipeline will open a new avenue for low-resource speech recognition and significantly facilitate research in this area.

Ziyuan Wang, Feiming Yang, Rui Fan

Graph neural networks have achieved state-of-the-art accuracy for graph node classification. However, GNNs are difficult to scale to large graphs, for example frequently encountering out-of-memory errors on even moderate size graphs. Recent works have sought to address this problem using a two-stage approach, which first aggregates data along graph edges, then trains a classifier without using additional graph information. These methods can run on much larger graphs and are orders of magnitude faster than GNNs, but achieve lower classification accuracy. We propose a novel two-stage algorithm based on a simple but effective observation: we should first train a classifier then aggregate, rather than the other way around. We show our algorithm is faster and can handle larger graphs than existing two-stage algorithms, while achieving comparable or higher accuracy than popular GNNs. We also present a theoretical basis to explain our algorithm's improved accuracy, by giving a synthetic nonlinear dataset in which performing aggregation before classification actually decreases accuracy compared to doing classification alone, while our classify then aggregate approach substantially improves accuracy compared to classification alone.

He Wang, Yifei Shen, Ziyuan Wang et al.

In this paper, we investigate the decentralized statistical inference problem, where a network of agents cooperatively recover a (structured) vector from private noisy samples without centralized coordination. Existing optimization-based algorithms suffer from issues of model mismatch and poor convergence speed, and thus their performance would be degraded, provided that the number of communication rounds is limited. This motivates us to propose a learning-based framework, which unrolls well-noted decentralized optimization algorithms (e.g., Prox-DGD and PG-EXTRA) into graph neural networks (GNNs). By minimizing the recovery error via end-to-end training, this learning-based framework resolves the model mismatch issue. Our convergence analysis (with PG-EXTRA as the base algorithm) reveals that the learned model parameters may accelerate the convergence and reduce the recovery error to a large extent. The simulation results demonstrate that the proposed GNN-based learning methods prominently outperform several state-of-the-art optimization-based algorithms in convergence speed and recovery error.

Peng Zhang, Yang Wang, Xutong Liu et al.

Background. Many mutation reduction strategies, which aim to reduce the number of mutants, have been proposed. Problem. It is important to measure the ability of a mutation reduction strategy to maintain test suite effectiveness evaluation. However, existing evaluation indicators are unable to measure the "order-preserving ability". Objective. We aim to propose evaluation indicators to measure the "order-preserving ability" of a mutation reduction strategy, which is important but missing in our community. Method. Given a test suite on a Software Under Test (SUT) with a set of original mutants, we leverage the test suite to generate a group of test suites that have a partial order relationship in fault detecting potential. When evaluating a reduction strategy, we first construct two partial order relationships among the generated test suites in terms of mutation score, one with the original mutants and another with the reduced mutants. Then, we measure the extent to which the two partial order relationships are consistent. The more consistent the two partial order relationships are, the stronger the Order Preservation (OP) of the mutation reduction strategy is, and the more effective the reduction strategy is. Furthermore, we propose Effort-aware Relative Order Preservation (EROP) to measure how much gain a mutation reduction strategy can provide compared with a random reduction strategy. Result. The experimental results show that OP and EROP are able to efficiently measure the "order-preserving ability" of a mutation reduction strategy. As a result, they have a better ability to distinguish various mutation reduction strategies compared with the existing evaluation indicators. Conclusion. We suggest, for the researchers, that OP and EROP should be used to measure the effectiveness of a mutant reduction strategy.

Tiexin Qin, Ziyuan Wang, Kelei He et al.

Conventional data augmentation realized by performing simple pre-processing operations (\eg, rotation, crop, \etc) has been validated for its advantage in enhancing the performance for medical image segmentation. However, the data generated by these conventional augmentation methods are random and sometimes harmful to the subsequent segmentation. In this paper, we developed a novel automatic learning-based data augmentation method for medical image segmentation which models the augmentation task as a trial-and-error procedure using deep reinforcement learning (DRL). In our method, we innovatively combine the data augmentation module and the subsequent segmentation module in an end-to-end training manner with a consistent loss. Specifically, the best sequential combination of different basic operations is automatically learned by directly maximizing the performance improvement (\ie, Dice ratio) on the available validation set. We extensively evaluated our method on CT kidney tumor segmentation which validated the promising results of our method.

Marco Virgolin, Ziyuan Wang, Tanja Alderliesten et al.

Machine Learning (ML) is proving extremely beneficial in many healthcare applications. In pediatric oncology, retrospective studies that investigate the relationship between treatment and late adverse effects still rely on simple heuristics. To assess the effects of radiation therapy, treatment plans are typically simulated on phantoms, i.e., virtual surrogates of patient anatomy. Currently, phantoms are built according to reasonable, yet simple, human-designed criteria. This often results in a lack of individualization. We present a novel approach that combines imaging and ML to build individualized phantoms automatically. Given the features of a patient treated historically (only 2D radiographs available), and a database of 3D Computed Tomography (CT) imaging with organ segmentations and relative patient features, our approach uses ML to predict how to assemble a patient-specific phantom automatically. Experiments on 60 abdominal CTs of pediatric patients show that our approach constructs significantly more representative phantoms than using current phantom building criteria, in terms of location and shape of the abdomen and of two considered organs, the liver and the spleen. Among several ML algorithms considered, the Gene-pool Optimal Mixing Evolutionary Algorithm for Genetic Programming (GP-GOMEA) is found to deliver the best performing models, which are, moreover, transparent and interpretable mathematical expressions.

Tianxing He, Shengcheng Yu, Ziyuan Wang et al.

Nowadays, people strive to improve the accuracy of deep learning models. However, very little work has focused on the quality of data sets. In fact, data quality determines model quality. Therefore, it is important for us to make research on how data quality affects on model quality. In this paper, we mainly consider four aspects of data quality, including Dataset Equilibrium, Dataset Size, Quality of Label, Dataset Contamination. We deign experiment on MNIST and Cifar-10 and try to find out the influence the four aspects make on model quality. Experimental results show that four aspects all have decisive impact on the quality of models. It means that decrease in data quality in these aspects will reduce the accuracy of model.