The growing capabilities of Artificial Intelligence (AI), particularly Large Language Models (LLMs), prompt a reassessment of the interaction mechanisms between users and their devices. Currently, users are required to use a set of high-level applications to achieve their desired results. However, the advent of AI may signal a shift in this regard, as its capabilities have generated novel prospects for user-provided intent resolution through the deployment of model-generated code. This development represents a significant progression in the realm of hybrid workflows, where human and artificial intelligence collaborate to address user intentions, with the former responsible for defining these intentions and the latter for implementing the solutions to address them. In this paper, we investigate the feasibility of generating and executing workflows through code generation that results from prompting an LLM with a concrete user intention, and a simplified application programming interface for a GUI-less operating system. We provide an in-depth analysis and comparison of various user intentions, the resulting code, and its execution. The findings demonstrate the general feasibility of our approach and that the employed LLM, GPT-4o-mini, exhibits remarkable proficiency in the generation of code-oriented workflows in accordance with provided user intentions.
Large Language Model (LLM) Agents have demonstrated remarkable capabilities in task automation and intelligent decision-making, driving the widespread adoption of agent development frameworks such as LangChain and AutoGen. However, these frameworks predominantly serve developers with extensive technical expertise - a significant limitation considering that only 0.03 % of the global population possesses the necessary programming skills. This stark accessibility gap raises a fundamental question: Can we enable everyone, regardless of technical background, to build their own LLM agents using natural language alone? To address this challenge, we introduce AutoAgent-a Fully-Automated and highly Self-Developing framework that enables users to create and deploy LLM agents through Natural Language Alone. Operating as an autonomous Agent Operating System, AutoAgent comprises four key components: i) Agentic System Utilities, ii) LLM-powered Actionable Engine, iii) Self-Managing File System, and iv) Self-Play Agent Customization module. This lightweight yet powerful system enables efficient and dynamic creation and modification of tools, agents, and workflows without coding requirements or manual intervention. Beyond its code-free agent development capabilities, AutoAgent also serves as a versatile multi-agent system for General AI Assistants. Comprehensive evaluations on the GAIA benchmark demonstrate AutoAgent's effectiveness in generalist multi-agent tasks, surpassing existing state-of-the-art methods. Furthermore, AutoAgent's Retrieval-Augmented Generation (RAG)-related capabilities have shown consistently superior performance compared to many alternative LLM-based solutions.
Activation functions are fundamental elements of deep learning architectures as they significantly influence training dynamics. ReLU, while widely used, is prone to the dying neuron problem, which has been mitigated by variants such as LeakyReLU, PReLU, and ELU that better handle negative neuron outputs. Recently, self-gated activations like GELU and Swish have emerged as state-of-the-art alternatives, leveraging their smoothness to ensure stable gradient flow and prevent neuron inactivity. In this work, we introduce the Gompertz Linear Unit (GoLU), a novel self-gated activation function defined as $\mathrm{GoLU}(x) = x \, \mathrm{Gompertz}(x)$, where $\mathrm{Gompertz}(x) = e^{-e^{-x}}$. The GoLU activation leverages the right-skewed asymmetry in the Gompertz function to reduce variance in the latent space more effectively compared to GELU and Swish, while preserving robust gradient flow. Extensive experiments across diverse tasks, including Image Classification, Language Modeling, Semantic Segmentation, Object Detection, Instance Segmentation, and Diffusion, highlight GoLU's superior performance relative to state-of-the-art activation functions, establishing GoLU as a robust alternative to existing activation functions.
Knowledge editing is a promising way to improve factuality in large language models, but recent studies have shown significant model degradation during sequential editing. In this paper, we formalize the popular locate-then-edit methods as a two-step fine-tuning process, allowing us to precisely identify the root cause of this degradation. We show that model degradation occurs due to (1) over-optimization of internal activations and (2) continuous norm-growth of edited matrices. To mitigate these issues, we introduce two regularization techniques: (1) Most-Probable Early Stopping (MPES) and (2) explicit Frobenius norm-constraint. We demonstrate that applying these simple yet effective regularization techniques at key points in the editing process can substantially mitigate model degradation. Combining these regularization methods enables scaling locate-then-edit methods to 10,000 edits while reducing editing time by 42-61%. These results show that targeted regularization is essential for lifelong knowledge editing.
Weather forecasting is essential but remains computationally intensive and physically incomplete in traditional numerical weather prediction (NWP) methods. Deep learning (DL) models offer efficiency and accuracy but often ignore physical laws, limiting interpretability and generalization. We propose PhyDL-NWP, a physics-guided deep learning framework that integrates physical equations with latent force parameterization into data-driven models. It predicts weather variables from arbitrary spatiotemporal coordinates, computes physical terms via automatic differentiation, and uses a physics-informed loss to align predictions with governing dynamics. PhyDL-NWP enables resolution-free downscaling by modeling weather as a continuous function and fine-tunes pre-trained models with minimal overhead, achieving up to 170x faster inference with only 55K parameters. Experiments show that PhyDL-NWP improves both forecasting performance and physical consistency.
Query routing, the task to route user queries to different large language model (LLM) endpoints, can be considered as a text classification problem. However, out-of-distribution queries must be handled properly, as those could be questions about unrelated domains, queries in other languages, or even contain unsafe text. Here, we thus study a \emph{guarded} query routing problem, for which we first introduce the Guarded Query Routing Benchmark (GQR-Bench), which covers three exemplary target domains (law, finance, and healthcare), and seven datasets to test robustness against out-of-distribution queries. We then use GQR-Bench to contrast the effectiveness and efficiency of LLM-based routing mechanisms (GPT-4o-mini, Llama-3.2-3B, and Llama-3.1-8B), standard LLM-based guardrail approaches (LlamaGuard and NVIDIA NeMo Guardrails), continuous bag-of-words classifiers (WideMLP, fastText), and traditional machine learning models (SVM, XGBoost). Our results show that WideMLP, enhanced with out-of-domain detection capabilities, yields the best trade-off between accuracy (88\%) and speed (<4ms). The embedding-based fastText excels at speed (<1ms) with acceptable accuracy (80\%), whereas LLMs yield the highest accuracy (91\%) but are comparatively slow (62ms for local Llama-3.1:8B and 669ms for remote GPT-4o-mini calls). Our findings challenge the automatic reliance on LLMs for (guarded) query routing and provide concrete recommendations for practical applications. GQR-Bench will be released as a Python package -- \texttt{gqr}.
Transformers, the standard implementation for large language models (LLMs), typically consist of tens to hundreds of discrete layers. While more layers can lead to better performance, this approach has been challenged as far from efficient, especially given the superiority of continuous layers demonstrated by diffusion and flow-based models for image generation. We propose the Latent Flow Transformer (LFT), which replaces a block of layers with a single learned transport operator trained via flow matching, offering significant compression while maintaining compatibility with the original architecture. Additionally, we address the limitations of existing flow-based methods in \textit{preserving coupling} by introducing the Flow Walking (FW) algorithm. On the Pythia-410M model, LFT trained with flow matching compresses 6 of 24 layers and outperforms directly skipping 2 layers (KL Divergence of LM logits at 0.407 vs. 0.529), demonstrating the feasibility of this design. When trained with FW, LFT further distills 12 layers into one while reducing the KL to 0.736 surpassing that from skipping 3 layers (0.932), significantly narrowing the gap between autoregressive and flow-based generation paradigms.
Process Reward Models (PRMs) have demonstrated promising results in mathematical reasoning, but existing process annotation approaches, whether through human annotations or Monte Carlo simulations, remain computationally expensive. In this paper, we introduce Step COmpression for Process Estimation (SCOPE), a novel compression-based approach that significantly reduces annotation costs. We first translate natural language reasoning steps into code and normalize them through Abstract Syntax Tree, then merge equivalent steps to construct a prefix tree. Unlike simulation-based methods that waste numerous samples on estimation, SCOPE leverages a compression-based prefix tree where each root-to-leaf path serves as a training sample, reducing the complexity from $O(NMK)$ to $O(N)$. We construct a large-scale dataset containing 196K samples with only 5% of the computational resources required by previous methods. Empirical results demonstrate that PRMs trained on our dataset consistently outperform existing automated annotation approaches on both Best-of-N strategy and ProcessBench.
Function approximation is a critical task in various fields. However, existing neural network approaches struggle with locally complex or discontinuous functions due to their reliance on a single global model covering the entire problem space. We propose X-KAN, a novel method that optimizes multiple local Kolmogorov-Arnold Networks (KANs) through an evolutionary rule-based machine learning framework called XCSF. X-KAN combines KAN's high expressiveness with XCSF's adaptive partitioning capability by implementing local KAN models as rule consequents and defining local regions via rule antecedents. Our experimental results on artificial test functions and real-world datasets demonstrate that X-KAN significantly outperforms conventional methods, including XCSF, Multi-Layer Perceptron, and KAN, in terms of approximation accuracy. Notably, X-KAN effectively handles functions with locally complex or discontinuous structures that are challenging for conventional KAN, using a compact set of rules (average 7.2 $\pm$ 2.3 rules). These results validate the effectiveness of using KAN as a local model in XCSF, which evaluates the rule fitness based on both accuracy and generality. Our X-KAN implementation is available at https://github.com/YNU-NakataLab/X-KAN.
Mathematical modeling is a cornerstone of scientific discovery and engineering practice, enabling the translation of real-world problems into formal systems across domains such as physics, biology, and economics. Unlike mathematical reasoning, which assumes a predefined formulation, modeling requires open-ended problem analysis, abstraction, and principled formalization. While Large Language Models (LLMs) have shown strong reasoning capabilities, they fall short in rigorous model construction, limiting their utility in real-world problem-solving. To this end, we formalize the task of LLM-powered real-world mathematical modeling, where agents must analyze problems, construct domain-appropriate formulations, and generate complete end-to-end solutions. We introduce MM-Bench, a curated benchmark of 111 problems from the Mathematical Contest in Modeling (MCM/ICM), spanning the years 2000 to 2025 and across ten diverse domains such as physics, biology, and economics. To tackle this task, we propose MM-Agent, an expert-inspired framework that decomposes mathematical modeling into four stages: open-ended problem analysis, structured model formulation, computational problem solving, and report generation. Experiments on MM-Bench show that MM-Agent significantly outperforms baseline agents, achieving an 11.88\% improvement over human expert solutions while requiring only 15 minutes and \$0.88 per task using GPT-4o. Furthermore, under official MCM/ICM protocols, MM-Agent assisted two undergraduate teams in winning the Finalist Award (\textbf{top 2.0\% among 27,456 teams}) in MCM/ICM 2025, demonstrating its practical effectiveness as a modeling copilot. Our code is available at https://github.com/usail-hkust/LLM-MM-Agent
Spatial labeling assigns labels to specific spatial locations to characterize their spatial properties and relationships, with broad applications in scientific research and practice. Measuring the similarity between two spatial labelings is essential for understanding their differences and the contributing factors, such as changes in location properties or labeling methods. An adequate and unbiased measurement of spatial labeling similarity should consider the number of matched labels (label agreement), the topology of spatial label distribution, and the heterogeneous impacts of mismatched labels. However, existing methods often fail to account for all these aspects. To address this gap, we propose a methodological framework to guide the development of methods that meet these requirements. Given two spatial labelings, the framework transforms them into graphs based on location organization, labels, and attributes (e.g., location significance). The distributions of their graph attributes are then extracted, enabling an efficient computation of distributional discrepancy to reflect the dissimilarity level between the two labelings. We further provide a concrete implementation of this framework, termed Spatial Labeling Analogy Metric (SLAM), along with an analysis of its theoretical foundation, for evaluating spatial labeling results in spatial transcriptomics (ST) \textit{as per} their similarity with ground truth labeling. Through a series of carefully designed experimental cases involving both simulated and real ST data, we demonstrate that SLAM provides a comprehensive and accurate reflection of labeling quality compared to other well-established evaluation metrics. Our code is available at https://github.com/YihDu/SLAM.
Despite advances in deep learning for education, student knowledge tracing and behavior modeling face persistent challenges: limited personalization, inadequate modeling of diverse learning activities (especially non-assessed materials), and overlooking the interplay between knowledge acquisition and behavioral patterns. Practical limitations, such as fixed-size sequence segmentation, frequently lead to the loss of contextual information vital for personalized learning. Moreover, reliance on student performance on assessed materials limits the modeling scope, excluding non-assessed interactions like lectures. To overcome these shortcomings, we propose Knowledge Modeling and Material Prediction (KMaP), a stateful multi-task approach designed for personalized and simultaneous modeling of student knowledge and behavior. KMaP employs clustering-based student profiling to create personalized student representations, improving predictions of future learning resource preferences. Extensive experiments on two real-world datasets confirm significant behavioral differences across student clusters and validate the efficacy of the KMaP model.
To enhance the reliability and credibility of graph neural networks (GNNs) and improve the transparency of their decision logic, a new field of explainability of GNNs (XGNN) has emerged. However, two major limitations severely degrade the performance and hinder the generalizability of existing XGNN methods: they (a) fail to capture the complete decision logic of GNNs across diverse distributions in the entire dataset's sample space, and (b) impose strict prerequisites on edge properties and GNN internal accessibility. To address these limitations, we propose OPEN, a novel c\textbf{O}mprehensive and \textbf{P}rerequisite-free \textbf{E}xplainer for G\textbf{N}Ns. OPEN, as the first work in the literature, can infer and partition the entire dataset's sample space into multiple environments, each containing graphs that follow a distinct distribution. OPEN further learns the decision logic of GNNs across different distributions by sampling subgraphs from each environment and analyzing their predictions, thus eliminating the need for strict prerequisites. Experimental results demonstrate that OPEN captures nearly complete decision logic of GNNs, outperforms state-of-the-art methods in fidelity while maintaining similar efficiency, and enhances robustness in real-world scenarios.
Deep learning models often achieve high performance by inadvertently learning spurious correlations between targets and non-essential features. For example, an image classifier may identify an object via its background that spuriously correlates with it. This prediction behavior, known as spurious bias, severely degrades model performance on data that lacks the learned spurious correlations. Existing methods on spurious bias mitigation typically require a variety of data groups with spurious correlation annotations called group labels. However, group labels require costly human annotations and often fail to capture subtle spurious biases such as relying on specific pixels for predictions. In this paper, we propose a novel post hoc spurious bias mitigation framework without requiring group labels. Our framework, termed ShortcutProbe, identifies prediction shortcuts that reflect potential non-robustness in predictions in a given model's latent space. The model is then retrained to be invariant to the identified prediction shortcuts for improved robustness. We theoretically analyze the effectiveness of the framework and empirically demonstrate that it is an efficient and practical tool for improving a model's robustness to spurious bias on diverse datasets.
Numerous works have noted significant similarities in how machine learning models represent the world, even across modalities. Although much effort has been devoted to uncovering properties and metrics on which these models align, surprisingly little work has explored causes of this similarity. To advance this line of inquiry, this work explores how two possible causal factors -- dataset overlap and task overlap -- influence downstream model similarity. The exploration of dataset overlap is motivated by the reality that large-scale generative AI models are often trained on overlapping datasets of scraped internet data, while the exploration of task overlap seeks to substantiate claims from a recent work, the Platonic Representation Hypothesis, that task similarity may drive model similarity. We evaluate the effects of both factors through a broad set of experiments. We find that both positively correlate with higher representational similarity and that combining them provides the strongest effect. Our code and dataset are published.
Modern LLMs are increasingly deep, and depth correlates with performance, albeit with diminishing returns. However, do these models use their depth efficiently? Do they compose more features to create higher-order computations that are impossible in shallow models, or do they merely spread the same kinds of computation out over more layers? To address these questions, we analyze the residual stream of the Llama 3.1 and Qwen 3 family of models. We find: First, comparing the output of the sublayers to the residual stream reveals that layers in the second half contribute much less than those in the first half, with a clear phase transition between the two halves. Second, skipping layers in the second half has a much smaller effect on future computations and output predictions. Third, for multihop tasks, we are unable to find evidence that models are using increased depth to compose subresults in examples involving many hops. Fourth, we seek to directly address whether deeper models are using their additional layers to perform new kinds of computation. To do this, we train linear maps from the residual stream of a shallow model to a deeper one. We find that layers with the same relative depth map best to each other, suggesting that the larger model simply spreads the same computations out over its many layers. All this evidence suggests that deeper models are not using their depth to learn new kinds of computation, but only using the greater depth to perform more fine-grained adjustments to the residual. This may help explain why increasing scale leads to diminishing returns for stacked Transformer architectures.
The past decades witness the significant advancements in time series forecasting (TSF) across various real-world domains, including e-commerce and disease spread prediction. However, TSF is usually constrained by the uncertainty dilemma of predicting future data with limited past observations. To settle this question, we explore the use of Cross-Future Behavior (CFB) in TSF, which occurs before the current time but takes effect in the future. We leverage CFB features and propose the CRoss-Future Behavior Awareness based Time Series Forecasting method (CRAFT). The core idea of CRAFT is to utilize the trend of cross-future behavior to mine the trend of time series data to be predicted. Specifically, to settle the sparse and partial flaws of cross-future behavior, CRAFT employs the Koopman Predictor Module to extract the key trend and the Internal Trend Mining Module to supplement the unknown area of the cross-future behavior matrix. Then, we introduce the External Trend Guide Module with a hierarchical structure to acquire more representative trends from higher levels. Finally, we apply the demand-constrained loss to calibrate the distribution deviation of prediction results. We conduct experiments on real-world dataset. Experiments on both offline large-scale dataset and online A/B test demonstrate the effectiveness of CRAFT. Our dataset and code is available at https://github.com/CRAFTinTSF/CRAFT.
Modern video understanding systems excel at tasks such as scene classification, object detection, and short video retrieval. However, as video analysis becomes increasingly central to real-world applications, there is a growing need for proactive video agents for the systems that not only interpret video streams but also reason about events and take informed actions. A key obstacle in this direction is temporal reasoning: while deep learning models have made remarkable progress in recognizing patterns within individual frames or short clips, they struggle to understand the sequencing and dependencies of events over time, which is critical for action-driven decision-making. Addressing this limitation demands moving beyond conventional deep learning approaches. We posit that tackling this challenge requires a neuro-symbolic perspective, where video queries are decomposed into atomic events, structured into coherent sequences, and validated against temporal constraints. Such an approach can enhance interpretability, enable structured reasoning, and provide stronger guarantees on system behavior, all key properties for advancing trustworthy video agents. To this end, we present a grand challenge to the research community: developing the next generation of intelligent video agents that integrate three core capabilities: (1) autonomous video search and analysis, (2) seamless real-world interaction, and (3) advanced content generation. By addressing these pillars, we can transition from passive perception to intelligent video agents that reason, predict, and act, pushing the boundaries of video understanding.
As AI systems become more capable, deceptive behaviors can undermine evaluation and mislead users at deployment. Recent work has shown that lie detectors can accurately classify deceptive behavior, but they are not typically used in the training pipeline due to concerns around contamination and objective hacking. We examine these concerns by incorporating a lie detector into the labelling step of LLM post-training and evaluating whether the learned policy is genuinely more honest, or instead learns to fool the lie detector while remaining deceptive. Using DolusChat, a novel 65k-example dataset with paired truthful/deceptive responses, we identify three key factors that determine the honesty of learned policies: amount of exploration during preference learning, lie detector accuracy, and KL regularization strength. We find that preference learning with lie detectors and GRPO can lead to policies which evade lie detectors, with deception rates of over 85\%. However, if the lie detector true positive rate (TPR) or KL regularization is sufficiently high, GRPO learns honest policies. In contrast, off-policy algorithms (DPO) consistently lead to deception rates under 25\% for realistic TPRs. Our results illustrate a more complex picture than previously assumed: depending on the context, lie-detector-enhanced training can be a powerful tool for scalable oversight, or a counterproductive method encouraging undetectable misalignment.
As post-training techniques evolve, large language models (LLMs) are increasingly augmented with structured multi-step reasoning abilities, often optimized through reinforcement learning. These reasoning-enhanced models outperform standard LLMs on complex tasks and now underpin many commercial LLM APIs. However, to protect proprietary behavior and reduce verbosity, providers typically conceal the reasoning traces while returning only the final answer. This opacity introduces a critical transparency gap: users are billed for invisible reasoning tokens, which often account for the majority of the cost, yet have no means to verify their authenticity. This opens the door to token count inflation, where providers may overreport token usage or inject synthetic, low-effort tokens to inflate charges. To address this issue, we propose CoIn, a verification framework that audits both the quantity and semantic validity of hidden tokens. CoIn constructs a verifiable hash tree from token embedding fingerprints to check token counts, and uses embedding-based relevance matching to detect fabricated reasoning content. Experiments demonstrate that CoIn, when deployed as a trusted third-party auditor, can effectively detect token count inflation with a success rate reaching up to 94.7%, showing the strong ability to restore billing transparency in opaque LLM services. The dataset and code are available at https://github.com/CASE-Lab-UMD/LLM-Auditing-CoIn.