arXiv subjects

Haotian Xia

Publications and source records attributed to Haotian Xia.

At least 19 recordsLinked to original sources

arXiv2026

SportD: How do VLMs physically strategize?

Vision-language models (VLMs) can describe a scene, but can they act well within one? We study whether VLMs can make sound strategic decisions, using soccer as an objective testbed with quantifiably-valued actions. We introduce SportD, a dataset and evaluation consisting of 1421 decision scenarios across professional men's and women's soccer games, where a VLM must decide what action to take next. Models on average select the optimal action around 27% of the time, less often than the professional players, and capture markedly less of the value at stake. Furthermore, they exhibit a clear preference for safer actions, favoring lower-variance, lower-value choices that also make less physical progress toward goal. Frontier VLMs are better at estimating whether an action will succeed, placing the highest-success-probability action among their top choices in 72-85% of cases. Yet VLMs systematically conflate likelihood with value, assigning higher value to actions that are more likely to succeed ($ρ=+0.30$ to $+0.52$), despite no such relationship in the ground truth ($ρ=-0.08$). Modifying the deliberation instructions to encourage risk-taking brings the frontier models closer to the players' skill levels. SportD opens a new direction for rigorously evaluating physical strategic decision-making in VLMs, showing that careful decomposition of their choices can reveal the mechanisms underlying systematic biases such as risk aversion.

cs.AI↗

arXiv2026

Can LLM-as-a-Judge Reliably Verify Rubrics in Agentic Scenarios?

Rubric-based scoring has become a widely used paradigm in model evaluation, typically with LLM-as-a-Judge (LaaJ) for rubric scoring. However, the reliability of LaaJ for rubric scoring remains underexplored. This concern is especially pronounced in agentic scenarios, where long, complex outputs further challenge reliable scoring. To address this, we conduct a systematic meta-evaluation of LaaJ reliability for rubric verification. We introduce RuVerBench, the first benchmark for assessing LaaJ reliability in rubric verification for agentic scenarios. RuVerBench covers two prevalent agentic domains, deep research and agentic coding, with 2,458 instances, each containing a model-generated output, a rubric, and a human-annotated label indicating whether the output satisfies the rubric. Using RuVerBench, we evaluate numerous frontier LLMs and find that even the most advanced models achieve strong performance but still exhibit substantial noise. We further analyze the impact of key LaaJ strategies, including prompt design, batching, and majority voting, on rubric verification. We find that weaker models are more sensitive to prompt variations, batched verification presents a trade-off between accuracy and efficiency, and majority voting yields effective but diminishing returns. We have released our dataset and code to facilitate future research: https://github.com/THU-KEG/RuVerBench.

cs.CL↗

arXiv2026

Perception Before Reasoning: Dynamic Latent Reasoning for Video Understanding and Question Answering

Video question answering requires models to ground language queries in visual evidence and, when necessary, reason over that evidence across time. Existing methods typically rely on long textual chain-of-thought rationales, even though many questions can be answered as soon as the relevant object, action, or frame is localized. We propose Dynamic Latent Reasoning (DyLaR), which first grounds a question in a short block of perception latents (continuous hidden states that encode query-relevant visual evidence), and then adaptively decides whether to append reasoning latents (continuous thoughts that reason over this evidence in latent space) before answering. DyLaR learns this behavior by grounding perception latents in verified visual evidence and distilling verified rationales into reasoning latents, followed by reinforcement learning that further refines when to reason. Across nine video benchmarks and four multimodal language model backbones, DyLaR improves average accuracy over same-backbone baselines while generating fewer than 20 tokens per query. On Qwen3-VL-4B, for example, DyLaR improves average accuracy over Qwen3-VL-4B-Thinking from 54.0 to 58.2 while reducing response length from 1,220.7 to 18.5 tokens per query. Ablations further show that grounded perception latents, rationale-supervised reasoning latents, and adaptive routing each improve accuracy.

cs.CV↗

arXiv2026

StoryAlign: Evaluating and Training Reward Models for Story Generation

Story generation aims to automatically produce coherent, structured, and engaging narratives. Although large language models (LLMs) have significantly advanced text generation, stories generated by LLMs still diverge from human-authored works regarding complex narrative structure and human-aligned preferences. A key reason is the absence of effective modeling of human story preferences, which are inherently subjective and under-explored. In this work, we systematically evaluate the modeling of human story preferences and introduce StoryRMB, the first benchmark for assessing reward models on story preferences. StoryRMB contains $1,133$ high-quality, human-verified instances, each consisting of a prompt, one chosen story, and three rejected stories. We find existing reward models struggle to select human-preferred stories, with the best model achieving only $66.3\%$ accuracy. To address this limitation, we construct roughly $100,000$ high-quality story preference pairs across diverse domains and develop StoryReward, an advanced reward model for story preference trained on this dataset. StoryReward achieves state-of-the-art (SoTA) performance on StoryRMB, outperforming much larger models. We also adopt StoryReward in downstream test-time scaling applications for best-of-n (BoN) story selection and find that it generally chooses stories better aligned with human preferences. We will release our dataset, model, and code to facilitate future research. Related code and data are available at https://github.com/THU-KEG/StoryReward.

cs.CL↗

arXiv2026

DeepSport: A Multimodal Large Language Model for Comprehensive Sports Video Reasoning via Agentic Reinforcement Learning

Sports video understanding requires perceiving high-speed dynamics, complex rules, and long temporal contexts. Yet, current Multimodal Large Language Models (MLLMs) remain narrowly focused on single sports, specific tasks, or training-free paradigms. We introduce DeepSport, the first end-to-end trained MLLM for multi-task, multi-sport video understanding. DeepSport shifts from passive frame processing to active, iterative reasoning, dynamically extracting frames to "think with videos." To train our model, we curate a unified 78k-sample dataset via a rigorous three-step text-and-vision distillation pipeline. We then employ a progressive two-stage training strategy: a Sports Curriculum Supervised Fine-Tuning phase to build foundational perception, followed by Agentic Reinforcement Learning with a novel tool-use reward. Extensive experiments on a comprehensive 6.7k benchmark demonstrate that DeepSport achieves state-of-the-art performance, outperforming powerful proprietary and open-source models, while utilizing significantly fewer frames. Furthermore, it exhibits strong zero-shot transferability to unseen sports and broad motion recognition tasks, establishing a highly efficient and generalized foundation for complex video reasoning.

cs.CV↗

arXiv2026

SportR: A Benchmark for Multimodal Large Language Model Reasoning in Sports

Deeply understanding sports requires an intricate blend of fine-grained visual perception and rule-based reasoning - a challenge that pushes the limits of current multimodal models. To succeed, models must master three critical capabilities: perceiving nuanced visual details, applying abstract sport rule knowledge, and grounding that knowledge in specific visual evidence. Current sports benchmarks either cover single sports or lack the detailed reasoning chains and precise visual grounding needed to robustly evaluate these core capabilities in a multi-sport context. To address this gap, we introduce SportR, the first multi-sports large-scale benchmark designed to train and evaluate MLLMs on the fundamental reasoning required for sports intelligence. Our benchmark provides a dataset of 4,789 images and 2,052 videos. To enable granular evaluation, we structure our benchmark around a progressive hierarchy of question-answer pairs designed to probe reasoning at increasing depths - from simple infraction identification to complex penalty prediction. For the most advanced tasks requiring multi-step reasoning, such as determining penalties or explaining tactics, we provide 6,841 high-quality, human-authored Chain of Thought annotations. In addition, our benchmark incorporates both image and video modalities and provides manual bounding box annotations to test visual grounding in the image part directly. Extensive experiments demonstrate the profound difficulty of our benchmark. State-of-the-art baseline models perform poorly on our most challenging tasks. While training on our data via Supervised Fine-Tuning and Reinforcement Learning improves these scores, they remain relatively low, highlighting a significant gap in current model capabilities. SportR presents a new challenge for the community, providing a critical resource to drive future research in multimodal sports reasoning.

cs.CV↗

arXiv2025

Evaluating Hydro-Science and Engineering Knowledge of Large Language Models

Hydro-Science and Engineering (Hydro-SE) is a critical and irreplaceable domain that secures human water supply, generates clean hydropower energy, and mitigates flood and drought disasters. Featuring multiple engineering objectives, Hydro-SE is an inherently interdisciplinary domain that integrates scientific knowledge with engineering expertise. This integration necessitates extensive expert collaboration in decision-making, which poses difficulties for intelligence. With the rapid advancement of large language models (LLMs), their potential application in the Hydro-SE domain is being increasingly explored. However, the knowledge and application abilities of LLMs in Hydro-SE have not been sufficiently evaluated. To address this issue, we propose the Hydro-SE LLM evaluation benchmark (Hydro-SE Bench), which contains 4,000 multiple-choice questions. Hydro-SE Bench covers nine subfields and enables evaluation of LLMs in aspects of basic conceptual knowledge, engineering application ability, and reasoning and calculation ability. The evaluation results on Hydro-SE Bench show that the accuracy values vary among 0.74 to 0.80 for commercial LLMs, and among 0.41 to 0.68 for small-parameter LLMs. While LLMs perform well in subfields closely related to natural and physical sciences, they struggle with domain-specific knowledge such as industry standards and hydraulic structures. Model scaling mainly improves reasoning and calculation abilities, but there is still great potential for LLMs to better handle problems in practical engineering application. This study highlights the strengths and weaknesses of LLMs for Hydro-SE tasks, providing model developers with clear training targets and Hydro-SE researchers with practical guidance for applying LLMs.

cs.CL↗

arXiv2025

VisFinEval: A Scenario-Driven Chinese Multimodal Benchmark for Holistic Financial Understanding

Multimodal large language models (MLLMs) hold great promise for automating complex financial analysis. To comprehensively evaluate their capabilities, we introduce VisFinEval, the first large-scale Chinese benchmark that spans the full front-middle-back office lifecycle of financial tasks. VisFinEval comprises 15,848 annotated question-answer pairs drawn from eight common financial image modalities (e.g., K-line charts, financial statements, official seals), organized into three hierarchical scenario depths: Financial Knowledge & Data Analysis, Financial Analysis & Decision Support, and Financial Risk Control & Asset Optimization. We evaluate 21 state-of-the-art MLLMs in a zero-shot setting. The top model, Qwen-VL-max, achieves an overall accuracy of 76.3%, outperforming non-expert humans but trailing financial experts by over 14 percentage points. Our error analysis uncovers six recurring failure modes-including cross-modal misalignment, hallucinations, and lapses in business-process reasoning-that highlight critical avenues for future research. VisFinEval aims to accelerate the development of robust, domain-tailored MLLMs capable of seamlessly integrating textual and visual financial information. The data and the code are available at https://github.com/SUFE-AIFLM-Lab/VisFinEval.

cs.CE↗

arXiv2025

StoryWriter: A Multi-Agent Framework for Long Story Generation

Long story generation remains a challenge for existing large language models (LLMs), primarily due to two main factors: (1) discourse coherence, which requires plot consistency, logical coherence, and completeness in the long-form generation, and (2) narrative complexity, which requires an interwoven and engaging narrative. To address these challenges, we propose StoryWriter, a multi-agent story generation framework, which consists of three main modules: (1) outline agent, which generates event-based outlines containing rich event plots, character, and event-event relationships. (2) planning agent, which further details events and plans which events should be written in each chapter to maintain an interwoven and engaging story. (3) writing agent, which dynamically compresses the story history based on the current event to generate and reflect new plots, ensuring the coherence of the generated story. We conduct both human and automated evaluation, and StoryWriter significantly outperforms existing story generation baselines in both story quality and length. Furthermore, we use StoryWriter to generate a dataset, which contains about $6,000$ high-quality long stories, with an average length of $8,000$ words. We train the model Llama3.1-8B and GLM4-9B using supervised fine-tuning on LongStory and develop StoryWriter_GLM and StoryWriter_GLM, which demonstrates advanced performance in long story generation.

cs.CL↗

arXiv2025

SPORTU: A Comprehensive Sports Understanding Benchmark for Multimodal Large Language Models

Multimodal Large Language Models (MLLMs) are advancing the ability to reason about complex sports scenarios by integrating textual and visual information. To comprehensively evaluate their capabilities, we introduce SPORTU, a benchmark designed to assess MLLMs across multi-level sports reasoning tasks. SPORTU comprises two key components: SPORTU-text, featuring 900 multiple-choice questions with human-annotated explanations for rule comprehension and strategy understanding. This component focuses on testing models' ability to reason about sports solely through question-answering (QA), without requiring visual inputs; SPORTU-video, consisting of 1,701 slow-motion video clips across 7 different sports and 12,048 QA pairs, designed to assess multi-level reasoning, from simple sports recognition to complex tasks like foul detection and rule application. We evaluate four prevalent LLMs mainly utilizing few-shot learning paradigms supplemented by chain-of-thought (CoT) prompting on the SPORTU-text part. We evaluate four LLMs using few-shot learning and chain-of-thought (CoT) prompting on SPORTU-text. GPT-4o achieves the highest accuracy of 71%, but still falls short of human-level performance, highlighting room for improvement in rule comprehension and reasoning. The evaluation for the SPORTU-video part includes 7 proprietary and 6 open-source MLLMs. Experiments show that models fall short on hard tasks that require deep reasoning and rule-based understanding. Claude-3.5-Sonnet performs the best with only 52.6% accuracy on the hard task, showing large room for improvement. We hope that SPORTU will serve as a critical step toward evaluating models' capabilities in sports understanding and reasoning.

cs.CV↗

arXiv2024

FACTS: Fine-Grained Action Classification for Tactical Sports

Classifying fine-grained actions in fast-paced, close-combat sports such as fencing and boxing presents unique challenges due to the complexity, speed, and nuance of movements. Traditional methods reliant on pose estimation or fancy sensor data often struggle to capture these dynamics accurately. We introduce FACTS, a novel transformer-based approach for fine-grained action recognition that processes raw video data directly, eliminating the need for pose estimation and the use of cumbersome body markers and sensors. FACTS achieves state-of-the-art performance, with 90% accuracy on fencing actions and 83.25% on boxing actions. Additionally, we present a new publicly available dataset featuring 8 detailed fencing actions, addressing critical gaps in sports analytics resources. Our findings enhance training, performance analysis, and spectator engagement, setting a new benchmark for action classification in tactical sports.

cs.CV↗

arXiv2024

FinEval: A Chinese Financial Domain Knowledge Evaluation Benchmark for Large Language Models

Large language models have demonstrated outstanding performance in various natural language processing tasks, but their security capabilities in the financial domain have not been explored, and their performance on complex tasks like financial agent remains unknown. This paper presents FinEval, a benchmark designed to evaluate LLMs' financial domain knowledge and practical abilities. The dataset contains 8,351 questions categorized into four different key areas: Financial Academic Knowledge, Financial Industry Knowledge, Financial Security Knowledge, and Financial Agent. Financial Academic Knowledge comprises 4,661 multiple-choice questions spanning 34 subjects such as finance and economics. Financial Industry Knowledge contains 1,434 questions covering practical scenarios like investment research. Financial Security Knowledge assesses models through 1,640 questions on topics like application security and cryptography. Financial Agent evaluates tool usage and complex reasoning with 616 questions. FinEval has multiple evaluation settings, including zero-shot, five-shot with chain-of-thought, and assesses model performance using objective and subjective criteria. Our results show that Claude 3.5-Sonnet achieves the highest weighted average score of 72.9 across all financial domain categories under zero-shot setting. Our work provides a comprehensive benchmark closely aligned with Chinese financial domain.

cs.CL↗

arXiv2024

Sports Intelligence: Assessing the Sports Understanding Capabilities of Language Models through Question Answering from Text to Video

Understanding sports is crucial for the advancement of Natural Language Processing (NLP) due to its intricate and dynamic nature. Reasoning over complex sports scenarios has posed significant challenges to current NLP technologies which require advanced cognitive capabilities. Toward addressing the limitations of existing benchmarks on sports understanding in the NLP field, we extensively evaluated mainstream large language models for various sports tasks. Our evaluation spans from simple queries on basic rules and historical facts to complex, context-specific reasoning, leveraging strategies from zero-shot to few-shot learning, and chain-of-thought techniques. In addition to unimodal analysis, we further assessed the sports reasoning capabilities of mainstream video language models to bridge the gap in multimodal sports understanding benchmarking. Our findings highlighted the critical challenges of sports understanding for NLP. We proposed a new benchmark based on a comprehensive overview of existing sports datasets and provided extensive error analysis which we hope can help identify future research priorities in this field.

cs.CL↗

arXiv2024

SportQA: A Benchmark for Sports Understanding in Large Language Models

A deep understanding of sports, a field rich in strategic and dynamic content, is crucial for advancing Natural Language Processing (NLP). This holds particular significance in the context of evaluating and advancing Large Language Models (LLMs), given the existing gap in specialized benchmarks. To bridge this gap, we introduce SportQA, a novel benchmark specifically designed for evaluating LLMs in the context of sports understanding. SportQA encompasses over 70,000 multiple-choice questions across three distinct difficulty levels, each targeting different aspects of sports knowledge from basic historical facts to intricate, scenario-based reasoning tasks. We conducted a thorough evaluation of prevalent LLMs, mainly utilizing few-shot learning paradigms supplemented by chain-of-thought (CoT) prompting. Our results reveal that while LLMs exhibit competent performance in basic sports knowledge, they struggle with more complex, scenario-based sports reasoning, lagging behind human expertise. The introduction of SportQA marks a significant step forward in NLP, offering a tool for assessing and enhancing sports understanding in LLMs.

cs.CL↗

arXiv2024

Language and Multimodal Models in Sports: A Survey of Datasets and Applications

Recent integration of Natural Language Processing (NLP) and multimodal models has advanced the field of sports analytics. This survey presents a comprehensive review of the datasets and applications driving these innovations post-2020. We overviewed and categorized datasets into three primary types: language-based, multimodal, and convertible datasets. Language-based and multimodal datasets are for tasks involving text or multimodality (e.g., text, video, audio), respectively. Convertible datasets, initially single-modal (video), can be enriched with additional annotations, such as explanations of actions and video descriptions, to become multimodal, offering future potential for richer and more diverse applications. Our study highlights the contributions of these datasets to various applications, from improving fan experiences to supporting tactical analysis and medical diagnostics. We also discuss the challenges and future directions in dataset development, emphasizing the need for diverse, high-quality data to support real-time processing and personalized user experiences. This survey provides a foundational resource for researchers and practitioners aiming to leverage NLP and multimodal models in sports, offering insights into current trends and future opportunities in the field.

cs.CL↗

arXiv2024

VREN: Volleyball Rally Dataset with Expression Notation Language

This research is intended to accomplish two goals: The first goal is to curate a large and information rich dataset that contains crucial and succinct summaries on the players' actions and positions and the back-and-forth travel patterns of the volleyball in professional and NCAA Div-I indoor volleyball games. While several prior studies have aimed to create similar datasets for other sports (e.g. badminton and soccer), creating such a dataset for indoor volleyball is not yet realized. The second goal is to introduce a volleyball descriptive language to fully describe the rally processes in the games and apply the language to our dataset. Based on the curated dataset and our descriptive sports language, we introduce three tasks for automated volleyball action and tactic analysis using our dataset: (1) Volleyball Rally Prediction, aimed at predicting the outcome of a rally and helping players and coaches improve decision-making in practice, (2) Setting Type and Hitting Type Prediction, to help coaches and players prepare more effectively for the game, and (3) Volleyball Tactics and Attacking Zone Statistics, to provide advanced volleyball statistics and help coaches understand the game and opponent's tactics better. We conducted case studies to show how experimental results can provide insights to the volleyball analysis community. Furthermore, experimental evaluation based on real-world data establishes a baseline for future studies and applications of our dataset and language. This study bridges the gap between the indoor volleyball field and computer science. The dataset is available at: https://github.com/haotianxia/VREN.

cs.LG↗

arXiv2023

Advanced Volleyball Stats for All Levels: Automatic Setting Tactic Detection and Classification with a Single Camera

This paper presents PathFinder and PathFinderPlus, two novel end-to-end computer vision frameworks designed specifically for advanced setting strategy classification in volleyball matches from a single camera view. Our frameworks combine setting ball trajectory recognition with a novel set trajectory classifier to generate comprehensive and advanced statistical data. This approach offers a fresh perspective for in-game analysis and surpasses the current level of granularity in volleyball statistics. In comparison to existing methods used in our baseline PathFinder framework, our proposed ball trajectory detection methodology in PathFinderPlus exhibits superior performance for classifying setting tactics under various game conditions. This robustness is particularly advantageous in handling complex game situations and accommodating different camera angles. Additionally, our study introduces an innovative algorithm for automatic identification of the opposing team's right-side (opposite) hitter's current row (front or back) during gameplay, providing critical insights for tactical analysis. The successful demonstration of our single-camera system's feasibility and benefits makes high-level technical analysis accessible to volleyball enthusiasts of all skill levels and resource availability. Furthermore, the computational efficiency of our system allows for real-time deployment, enabling in-game strategy analysis and on-the-spot gameplan adjustments.

cs.CV↗

arXiv2023

Graph Encoding and Neural Network Approaches for Volleyball Analytics: From Game Outcome to Individual Play Predictions

This research aims to improve the accuracy of complex volleyball predictions and provide more meaningful insights to coaches and players. We introduce a specialized graph encoding technique to add additional contact-by-contact volleyball context to an already available volleyball dataset without any additional data gathering. We demonstrate the potential benefits of using graph neural networks (GNNs) on this enriched dataset for three different volleyball prediction tasks: rally outcome prediction, set location prediction, and hit type prediction. We compare the performance of our graph-based models to baseline models and analyze the results to better understand the underlying relationships in a volleyball rally. Our results show that the use of GNNs with our graph encoding yields a much more advanced analysis of the data, which noticeably improves prediction results overall. We also show that these baseline tasks can be significantly improved with simple adjustments, such as removing blocked hits. Lastly, we demonstrate the importance of choosing a model architecture that will better extract the important information for a certain task. Overall, our study showcases the potential strengths and weaknesses of using graph encodings in sports data analytics and hopefully will inspire future improvements in machine learning strategies across sports and applications by using graphbased encodings.

cs.LG↗