arXiv subjects

Haotian Wang

Publications and source records attributed to Haotian Wang.

At least 73 records · Page 4Linked to original sources

arXiv2025

PPA-Game: Characterizing and Learning Competitive Dynamics Among Online Content Creators

In this paper, we present the Proportional Payoff Allocation Game (PPA-Game), which characterizes situations where agents compete for divisible resources. In the PPA-game, agents select from available resources, and their payoffs are proportionately determined based on heterogeneous weights attributed to them. Such dynamics simulate content creators on online recommender systems like YouTube and TikTok, who compete for finite consumer attention, with content exposure reliant on inherent and distinct quality. We first conduct a game-theoretical analysis of the PPA-Game. While the PPA-Game does not always guarantee the existence of a pure Nash equilibrium (PNE), we identify prevalent scenarios ensuring its existence. Simulated experiments further prove that the cases where PNE does not exist rarely happen. Beyond analyzing static payoffs, we further discuss the agents' online learning about resource payoffs by integrating a multi-player multi-armed bandit framework. We propose an online algorithm facilitating each agent's maximization of cumulative payoffs over $T$ rounds. Theoretically, we establish that the regret of any agent is bounded by $O(\log^{1 + η} T)$ for any $η> 0$. Empirical results further validate the effectiveness of our online learning approach.

cs.GT↗

arXiv2025

BaseReward: A Strong Baseline for Multimodal Reward Model

The rapid advancement of Multimodal Large Language Models (MLLMs) has made aligning them with human preferences a critical challenge. Reward Models (RMs) are a core technology for achieving this goal, but a systematic guide for building state-of-the-art Multimodal Reward Models (MRMs) is currently lacking in both academia and industry. Through exhaustive experimental analysis, this paper aims to provide a clear ``recipe'' for constructing high-performance MRMs. We systematically investigate every crucial component in the MRM development pipeline, including \textit{reward modeling paradigms} (e.g., Naive-RM, Critic-based RM, and Generative RM), \textit{reward head architecture}, \textit{training strategies}, \textit{data curation} (covering over ten multimodal and text-only preference datasets), \textit{backbone model} and \textit{model scale}, and \textit{ensemble methods}. Based on these experimental insights, we introduce \textbf{BaseReward}, a powerful and efficient baseline for multimodal reward modeling. BaseReward adopts a simple yet effective architecture, built upon a {Qwen2.5-VL} backbone, featuring an optimized two-layer reward head, and is trained on a carefully curated mixture of high-quality multimodal and text-only preference data. Our results show that BaseReward establishes a new SOTA on major benchmarks such as MM-RLHF-Reward Bench, VL-Reward Bench, and Multimodal Reward Bench, outperforming previous models. Furthermore, to validate its practical utility beyond static benchmarks, we integrate BaseReward into a real-world reinforcement learning pipeline, successfully enhancing an MLLM's performance across various perception, reasoning, and conversational tasks. This work not only delivers a top-tier MRM but, more importantly, provides the community with a clear, empirically-backed guide for developing robust reward models for the next generation of MLLMs.

cs.CV↗

arXiv2025

Hierarchical Structure Sharing Empowers Multi-task Heterogeneous GNNs for Customer Expansion

Customer expansion, i.e., growing a business existing customer base by acquiring new customers, is critical for scaling operations and sustaining the long-term profitability of logistics companies. Although state-of-the-art works model this task as a single-node classification problem under a heterogeneous graph learning framework and achieve good performance, they struggle with extremely positive label sparsity issues in our scenario. Multi-task learning (MTL) offers a promising solution by introducing a correlated, label-rich task to enhance the label-sparse task prediction through knowledge sharing. However, existing MTL methods result in performance degradation because they fail to discriminate task-shared and task-specific structural patterns across tasks. This issue arises from their limited consideration of the inherently complex structure learning process of heterogeneous graph neural networks, which involves the multi-layer aggregation of multi-type relations. To address the challenge, we propose a Structure-Aware Hierarchical Information Sharing Framework (SrucHIS), which explicitly regulates structural information sharing across tasks in logistics customer expansion. SrucHIS breaks down the structure learning phase into multiple stages and introduces sharing mechanisms at each stage, effectively mitigating the influence of task-specific structural patterns during each stage. We evaluate StrucHIS on both private and public datasets, achieving a 51.41% average precision improvement on the private dataset and a 10.52% macro F1 gain on the public dataset. StrucHIS is further deployed at one of the largest logistics companies in China and demonstrates a 41.67% improvement in the success contract-signing rate over existing strategies, generating over 453K new orders within just two months.

cs.LG↗

arXiv2025

AgriEval: A Comprehensive Chinese Agricultural Benchmark for Large Language Models

In the agricultural domain, the deployment of large language models (LLMs) is hindered by the lack of training data and evaluation benchmarks. To mitigate this issue, we propose AgriEval, the first comprehensive Chinese agricultural benchmark with three main characteristics: (1) Comprehensive Capability Evaluation. AgriEval covers six major agriculture categories and 29 subcategories within agriculture, addressing four core cognitive scenarios: memorization, understanding, inference, and generation. (2) High-Quality Data. The dataset is curated from university-level examinations and assignments, providing a natural and robust benchmark for assessing the capacity of LLMs to apply knowledge and make expert-like decisions. (3) Diverse Formats and Extensive Scale. AgriEval comprises 14,697 multiple-choice questions and 2,167 open-ended question-and-answer questions, establishing it as the most extensive agricultural benchmark available to date. We also present comprehensive experimental results over 51 open-source and commercial LLMs. The experimental results reveal that most existing LLMs struggle to achieve 60% accuracy, underscoring the developmental potential in agricultural LLMs. Additionally, we conduct extensive experiments to investigate factors influencing model performance and propose strategies for enhancement. AgriEval is available at https://github.com/YanPioneer/AgriEval/.

cs.CL↗

arXiv2025

Inactive longitude and superflare in the active single-lined pre-main sequence binary V2279 Cyg

Young, solar-like stars in the pre-main sequence (PMS) stage exhibit vigorous magnetic activity that significantly influences their circumstellar environments and the processes of planetary formation and evolution. In binary systems, tidal forces and magnetic interactions can further shape the magnetic geometry. We report a longitudinal preference of star spots, chromospheric activities, and flares in the active single-lined spectroscopic PMS binary system V2279 Cyg, based on long-term photometric observations from \textit{Kepler} and \textit{TESS} alongside spectroscopic data from LAMOST. The system is classified as a weak-line T Tauri binary, with component masses estimated at 0.86 $M_\odot$ and 0.27 $M_\odot$. V2279 Cyg's nearly circular orbit is synchronized with its 4.126-day rotational period. Observations reveal large star spot regions clustered near the far-side hemisphere. Spectroscopic data show strong, double-peak H$α$ emission, the strength of which is highly correlated with star spot distribution, indicating the presence of an active longitude on the primary star. We also mapped the prominence structure co-rotating with the primary star, suggesting a dense structure close to the near-side hemisphere. Furthermore, we identify an inactive longitude of flares during the 4-year \textit{Kepler} observations, where the frequency of flare activity is significantly reduced after the superior conjunction, marking the first such identification in active binary systems. Additionally, a white light superflare, releasing energy of $2.5 \times 10^{37}$ erg, was detected in \textit{TESS} observations. These findings provide valuable insights into the magnetic field geometry and dynamo processes in PMS binaries, underscoring the critical role of tidal interactions in shaping magnetic activities.

astro-ph.SR↗

arXiv2025

PacGDC: Label-Efficient Generalizable Depth Completion with Projection Ambiguity and Consistency

Generalizable depth completion enables the acquisition of dense metric depth maps for unseen environments, offering robust perception capabilities for various downstream tasks. However, training such models typically requires large-scale datasets with metric depth labels, which are often labor-intensive to collect. This paper presents PacGDC, a label-efficient technique that enhances data diversity with minimal annotation effort for generalizable depth completion. PacGDC builds on novel insights into inherent ambiguities and consistencies in object shapes and positions during 2D-to-3D projection, allowing the synthesis of numerous pseudo geometries for the same visual scene. This process greatly broadens available geometries by manipulating scene scales of the corresponding depth maps. To leverage this property, we propose a new data synthesis pipeline that uses multiple depth foundation models as scale manipulators. These models robustly provide pseudo depth labels with varied scene scales, affecting both local objects and global layouts, while ensuring projection consistency that supports generalization. To further diversify geometries, we incorporate interpolation and relocation strategies, as well as unlabeled images, extending the data coverage beyond the individual use of foundation models. Extensive experiments show that PacGDC achieves remarkable generalizability across multiple benchmarks, excelling in diverse scene semantics/scales and depth sparsity/patterns under both zero-shot and few-shot settings. Code: https://github.com/Wang-xjtu/PacGDC.

cs.CV↗

arXiv2025

Physics-informed network paradigm with data generation and background noise removal for diverse distributed acoustic sensing applications

Distributed acoustic sensing (DAS) has attracted considerable attention across various fields and artificial intelligence (AI) technology plays an important role in DAS applications to realize event recognition and denoising. Existing AI models require real-world data (RWD), whether labeled or not, for training, which is contradictory to the fact of limited available event data in real-world scenarios. Here, a physics-informed DAS neural network paradigm is proposed, which does not need real-world events data for training. By physically modeling target events and the constraints of real world and DAS system, physical functions are derived to train a generative network for generation of DAS events data. DAS debackground net is trained by using the generated DAS events data to eliminate background noise in DAS data. The effectiveness of the proposed paradigm is verified in event identification application based on a public dataset of DAS spatiotemporal data and in belt conveyor fault monitoring application based on DAS time-frequency data, and achieved comparable or better performance than data-driven networks trained with RWD. Owing to the introduction of physical information and capability of background noise removal, the paradigm demonstrates generalization in same application on different sites. A fault diagnosis accuracy of 91.8% is achieved in belt conveyor field with networks which transferred from simulation test site without any fault events data of test site and field for training. The proposed paradigm is a prospective solution to address significant obstacles of data acquisition and intense noise in practical DAS applications and explore more potential fields for DAS.

cs.LG↗

arXiv2025

BLITZSCALE: Fast and Live Large Model Autoscaling with O(1) Host Caching

Model autoscaling is the key mechanism to achieve serverless model-as-a-service, but it faces a fundamental trade-off between scaling speed and storage/memory usage to cache parameters, and cannot meet frequent scaling requirements across multiple hosts. The key problem is that data plane performance is slow, and scaled instances remain stopped while parameters are loading. In this paper, we first show that the data plane can be made fast with no or O(1) caching by loading parameters through the compute network between GPUs because: (1) its speed is comparable to host cache and is underutilized, and (2) scaling multiple instances requires no or O(1) caching with network-optimized multicast. Second, autoscaling can be made live by breaking the scaling abstraction for inference from a coarse-grained instance-level to a fine-grained layer-level. This allows us to offload the layer computation from the overloaded serving instances to the scaled ones without waiting for the parameters to be fully loaded. Under real-world workloads, our system BLITZSCALE achieves up to 94 % lower tail latency reductions compared to state-of-the-art autoscaling system (ServerlessLLM), and it reduces the GPU time used for serving by 49 % when compared with serving systems that do not support autoscaling like DistServe and vLLM with the same service-level-agreement.

cs.DC↗

arXiv2025

UBench: Benchmarking Uncertainty in Large Language Models with Multiple Choice Questions

Despite recent progress in systematic evaluation frameworks, benchmarking the uncertainty of large language models (LLMs) remains a highly challenging task. Existing methods for benchmarking the uncertainty of LLMs face three key challenges: the need for internal model access, additional training, or high computational costs. This is particularly unfavorable for closed-source models. To this end, we introduce UBench, a new benchmark for evaluating the uncertainty of LLMs. Unlike other benchmarks, UBench is based on confidence intervals. It encompasses 11,978 multiple-choice questions spanning knowledge, language, understanding, and reasoning capabilities. Based on this, we conduct extensive experiments. This includes comparisons with other advanced uncertainty estimation methods, the assessment of the uncertainty of 20 LLMs, and an exploration of the effects of Chain-of-Thought (CoT) prompts, role-playing (RP) prompts, and temperature on model uncertainty. Our analysis reveals several crucial insights: 1) Our confidence interval-based methods are highly effective for uncertainty quantification; 2) Regarding uncertainty, outstanding open-source models show competitive performance versus closed-source models; 3) CoT and RP prompts present potential ways to improve model reliability, while the influence of temperature changes follows no universal rule. Our implementation is available at https://github.com/Cyno2232/UBENCH.

cs.CL↗

arXiv2025

AM-Thinking-v1: Advancing the Frontier of Reasoning at 32B Scale

We present AM-Thinking-v1, a 32B dense language model that advances the frontier of reasoning, embodying the collaborative spirit of open-source innovation. Outperforming DeepSeek-R1 and rivaling leading Mixture-of-Experts (MoE) models like Qwen3-235B-A22B and Seed1.5-Thinking, AM-Thinking-v1 achieves impressive scores of 85.3 on AIME 2024, 74.4 on AIME 2025, and 70.3 on LiveCodeBench, showcasing state-of-the-art mathematical and coding capabilities among open-source models of similar scale. Built entirely from the open-source Qwen2.5-32B base model and publicly available queries, AM-Thinking-v1 leverages a meticulously crafted post-training pipeline - combining supervised fine-tuning and reinforcement learning - to deliver exceptional reasoning capabilities. This work demonstrates that the open-source community can achieve high performance at the 32B scale, a practical sweet spot for deployment and fine-tuning. By striking a balance between top-tier performance and real-world usability, we hope AM-Thinking-v1 inspires further collaborative efforts to harness mid-scale models, pushing reasoning boundaries while keeping accessibility at the core of innovation. We have open-sourced our model on \href{https://huggingface.co/a-m-team/AM-Thinking-v1}{Hugging Face}.

cs.CL↗

arXiv2025

Not All Correct Answers Are Equal: Why Your Distillation Source Matters

Distillation has emerged as a practical and effective approach to enhance the reasoning capabilities of open-source language models. In this work, we conduct a large-scale empirical study on reasoning data distillation by collecting verified outputs from three state-of-the-art teacher models-AM-Thinking-v1, Qwen3-235B-A22B, and DeepSeek-R1-on a shared corpus of 1.89 million queries. We construct three parallel datasets and analyze their distributions, revealing that AM-Thinking-v1-distilled data exhibits greater token length diversity and lower perplexity. Student models trained on each dataset are evaluated on reasoning benchmarks including AIME2024, AIME2025, MATH500, and LiveCodeBench. The model distilled from AM-Thinking-v1 consistently achieves the best performance (e.g., 84.3 on AIME2024, 72.2 on AIME2025, 98.4 on MATH500, and 65.9 on LiveCodeBench) and demonstrates adaptive output behavior-producing longer responses for harder tasks and shorter ones for simpler tasks. These findings highlight the value of high-quality, verified reasoning traces. We release the AM-Thinking-v1 and Qwen3-235B-A22B distilled datasets to support future research on open and high-performing reasoning-oriented language models. The datasets are publicly available on Hugging Face\footnote{Datasets are available on Hugging Face: \href{https://huggingface.co/datasets/a-m-team/AM-Thinking-v1-Distilled}{AM-Thinking-v1-Distilled}, \href{https://huggingface.co/datasets/a-m-team/AM-Qwen3-Distilled}{AM-Qwen3-Distilled}.}.

cs.CL↗

arXiv2025

DeepDistill: Enhancing LLM Reasoning Capabilities via Large-Scale Difficulty-Graded Data Training

Although large language models (LLMs) have recently achieved remarkable performance on various complex reasoning benchmarks, the academic community still lacks an in-depth understanding of base model training processes and data quality. To address this, we construct a large-scale, difficulty-graded reasoning dataset containing approximately 3.34 million unique queries of varying difficulty levels and about 40 million distilled responses generated by multiple models over several passes. Leveraging pass rate and Coefficient of Variation (CV), we precisely select the most valuable training data to enhance reasoning capability. Notably, we observe a training pattern shift, indicating that reasoning-focused training based on base models requires higher learning rates for effective training. Using this carefully selected data, we significantly improve the reasoning capabilities of the base model, achieving a pass rate of 79.2\% on the AIME2024 mathematical reasoning benchmark. This result surpasses most current distilled models and closely approaches state-of-the-art performance. We provide detailed descriptions of our data processing, difficulty assessment, and training methodology, and have publicly released all datasets and methods to promote rapid progress in open-source long-reasoning LLMs. The dataset is available at: \href{https://huggingface.co/datasets/a-m-team/AM-DeepSeek-Distilled-40M}{https://huggingface.co/datasets/a-m-team/AM-DeepSeek-Distilled-40M}

cs.CL↗

arXiv2025

Exploring the Potential of Offline RL for Reasoning in LLMs: A Preliminary Study

Despite significant advances in long-context reasoning by large language models (LLMs), primarily through Online Reinforcement Learning (RL) methods, these approaches incur substantial computational costs and complexity. In contrast, simpler and more economical Offline RL methods remain underexplored. To address this gap, we investigate the effectiveness of Offline RL methods, specifically Direct Preference Optimization (DPO) and its length-desensitized variant LD-DPO, in enhancing the reasoning capabilities of LLMs. Extensive experiments across multiple reasoning benchmarks demonstrate that these simpler Offline RL methods substantially improve model performance, achieving an average enhancement of 3.3\%, with a particularly notable increase of 10.1\% on the challenging Arena-Hard benchmark. Furthermore, we analyze DPO's sensitivity to output length, emphasizing that increasing reasoning length should align with semantic richness, as indiscriminate lengthening may adversely affect model performance. We provide comprehensive descriptions of our data processing and training methodologies, offering empirical evidence and practical insights for developing more cost-effective Offline RL approaches.

cs.CL↗

arXiv2025

DeepSTA: A Spatial-Temporal Attention Network for Logistics Delivery Timely Rate Prediction in Anomaly Conditions

Prediction of couriers' delivery timely rates in advance is essential to the logistics industry, enabling companies to take preemptive measures to ensure the normal operation of delivery services. This becomes even more critical during anomaly conditions like the epidemic outbreak, during which couriers' delivery timely rate will decline markedly and fluctuates significantly. Existing studies pay less attention to the logistics scenario. Moreover, many works focusing on prediction tasks in anomaly scenarios fail to explicitly model abnormal events, e.g., treating external factors equally with other features, resulting in great information loss. Further, since some anomalous events occur infrequently, traditional data-driven methods perform poorly in these scenarios. To deal with them, we propose a deep spatial-temporal attention model, named DeepSTA. To be specific, to avoid information loss, we design an anomaly spatio-temporal learning module that employs a recurrent neural network to model incident information. Additionally, we utilize Node2vec to model correlations between road districts, and adopt graph neural networks and long short-term memory to capture the spatial-temporal dependencies of couriers. To tackle the issue of insufficient training data in abnormal circumstances, we propose an anomaly pattern attention module that adopts a memory network for couriers' anomaly feature patterns storage via attention mechanisms. The experiments on real-world logistics datasets during the COVID-19 outbreak in 2022 show the model outperforms the best baselines by 12.11% in MAE and 13.71% in MSE, demonstrating its superior performance over multiple competitive baselines.

cs.LG↗

arXiv2025

Learning to Estimate Package Delivery Time in Mixed Imbalanced Delivery and Pickup Logistics Services

Accurately estimating package delivery time is essential to the logistics industry, which enables reasonable work allocation and on-time service guarantee. This becomes even more necessary in mixed logistics scenarios where couriers handle a high volume of delivery and a smaller number of pickup simultaneously. However, most of the related works treat the pickup and delivery patterns on couriers' decision behavior equally, neglecting that the pickup has a greater impact on couriers' decision-making compared to the delivery due to its tighter time constraints. In such context, we have three main challenges: 1) multiple spatiotemporal factors are intricately interconnected, significantly affecting couriers' delivery behavior; 2) pickups have stricter time requirements but are limited in number, making it challenging to model their effects on couriers' delivery process; 3) couriers' spatial mobility patterns are critical determinants of their delivery behavior, but have been insufficiently explored. To deal with these, we propose TransPDT, a Transformer-based multi-task package delivery time prediction model. We first employ the Transformer encoder architecture to capture the spatio-temporal dependencies of couriers' historical travel routes and pending package sets. Then we design the pattern memory to learn the patterns of pickup in the imbalanced dataset via attention mechanism. We also set the route prediction as an auxiliary task of delivery time prediction, and incorporate the prior courier spatial movement regularities in prediction. Extensive experiments on real industry-scale datasets demonstrate the superiority of our method. A system based on TransPDT is deployed internally in JD Logistics to track more than 2000 couriers handling hundreds of thousands of packages per day in Beijing.

cs.LG↗

arXiv2025

Dual Boost-Driven Graph-Level Clustering Network

Graph-level clustering remains a pivotal yet formidable challenge in graph learning. Recently, the integration of deep learning with representation learning has demonstrated notable advancements, yielding performance enhancements to a certain degree. However, existing methods suffer from at least one of the following issues: 1. the original graph structure has noise, and 2. during feature propagation and pooling processes, noise is gradually aggregated into the graph-level embeddings through information propagation. Consequently, these two limitations mask clustering-friendly information, leading to suboptimal graph-level clustering performance. To this end, we propose a novel Dual Boost-Driven Graph-Level Clustering Network (DBGCN) to alternately promote graph-level clustering and filtering out interference information in a unified framework. Specifically, in the pooling step, we evaluate the contribution of features at the global and optimize them using a learnable transformation matrix to obtain high-quality graph-level representation, such that the model's reasoning capability can be improved. Moreover, to enable reliable graph-level clustering, we first identify and suppress information detrimental to clustering by evaluating similarities between graph-level representations, providing more accurate guidance for multi-view fusion. Extensive experiments demonstrated that DBGCN outperforms the state-of-the-art graph-level clustering methods on six benchmark datasets.

cs.LG↗

arXiv2025

Leveraging Reasoning Model Answers to Enhance Non-Reasoning Model Capability

Recent advancements in large language models (LLMs), such as DeepSeek-R1 and OpenAI-o1, have demonstrated the significant effectiveness of test-time scaling, achieving substantial performance gains across various benchmarks. These advanced models utilize deliberate "thinking" steps to systematically enhance answer quality. In this paper, we propose leveraging these high-quality outputs generated by reasoning-intensive models to improve less computationally demanding, non-reasoning models. We explore and compare methodologies for utilizing the answers produced by reasoning models to train and improve non-reasoning models. Through straightforward Supervised Fine-Tuning (SFT) experiments on established benchmarks, we demonstrate consistent improvements across various benchmarks, underscoring the potential of this approach for advancing the ability of models to answer questions directly.

cs.CL↗

arXiv2025

Initial Error Tolerant Distributed Mean Field Control under Partial and Discrete Information

In this paper, an initial error tolerant distributed mean field control method under partial and discrete information is introduced, where each agent only has discrete observations on its own state. First, we study agents' behavior in linear quadratic mean field games (LQMFGs) under heterogeneous erroneous information of the initial mean field state (MF-S), and formulate the relationships between initial errors and systemic deviations. Next, by capturing the initial error affection on the private trajectory of an agent, we give a distributed error estimation method based on maximum likelihood estimation (MLE), where each agent estimates information errors only based on discrete observations on its private trajectory. Furthermore, we establish an error-based segmented state estimation method, design the initial error tolerant distributed mean field control method (IET-DMFC), and demonstrate the consistent property of state estimation as observation frequency increases. Finally, simulations are performed to verify the efficiency of the algorithm and the consistent properties.

math.OC↗