arXiv subjects

Zhiqi Gao

Publications and source records attributed to Zhiqi Gao.

17 recordsLinked to original sources

arXiv2026

Models Can Model, But Can't Bind: Structured Grounding in Text-to-Optimization

Text-to-optimization requires two separable capabilities: modeling -- choosing the right optimization structure -- and binding -- grounding every coefficient, index, and parameter in the concrete problem data. We study this via Text2Opt-Bench, a scalable benchmark of solver-verified optimization problems spanning 12 categories, from textbook linear programs to stochastic and multi-objective formulations with up to thousands of variables. Across 10+ models, we find that accuracy collapses as instance data grows, even when the formulation itself is simple. We call this the effective binding limit. We study it with a family of techniques, BIND, that externalize numeric data to structured files so the model binds data programmatically rather than transcribing from the prompt. When using an oracle for externalizing data, we recover between 12 and 27 accuracy points, confirming binding as a key -- but recoverable -- failure mode. In a deployable setting without oracle access, we validate our hypothesis by finetuning a model exclusively on binding and show that it outperforms end-to-end SFT and RL across three structurally distinct optimization categories, with a 1.5B binding specialist alone matching a 7B end-to-end baseline.

cs.LG↗

arXiv2026

MedEasy: Designing AI Standardized Patients for Clinical Consultation Training

AI standardized patients are becoming a setting for professional training in clinical consultation. This paper presents MedEasy, a multi-agent system that organizes virtual-patient practice through patient dialogue, clinical actions, decision submission, documentation, and feedback. We first conducted a formative study with 12 clinical-year medical students through interviews and three co-design workshops. The findings informed a staged workflow, structured case records, action-contingent findings, and trajectory-based review. We then conducted an evaluative user study with a separate cohort of 12 clinical-year medical students, with each participant completing two counterbalanced cases. Learners interpreted MedEasy as a connected consultation environment. They used patient responses, examination findings, available actions, and feedback together to judge whether the represented case remained coherent. They valued repeatable practice and recorded review, while questioning missing actions and feedback criteria. The paper contributes design implications for AI-supported professional training systems that use case-specific standards to connect situated practice.

cs.HC↗

arXiv2026

Reading the Same Data Differently: Interpretive Labor Across System Boundaries in Electronic Monitoring

Electronic monitoring (EM) systems are increasingly used in community corrections to enforce spatial, temporal, and behavioral rules through continuous sensing. While prior work has examined EM as a criminal justice tool or as a mechanism for compliance, less is known about how sensed data become meaningful in everyday practice. This poster examines EM as a dual-sided sensing system in which supervised individuals and authorities reason about the same data stream from different positions. Based on semi-structured interviews with 26 supervised individuals and 12 authorities in China's community corrections system, we show that supervised individuals infer system logic from outcomes with limited visibility into how data are interpreted, while authorities reconstruct behavior from ambiguous traces using contextual knowledge, professional experience, and institutional procedures. We call this structural divergence interpretive misalignment. It emerges from asymmetric access to data, context, and reasoning processes, and it shapes behavior through probing, strategic adaptation, over-compliance, disengagement, and contestation. We contribute a CSCW account of continuous sensing as distributed interpretive work and identify design opportunities for making data-to-decision processes more legible, contestable, and accountable across system sides.

cs.HC↗

arXiv2026

Fine-Tuning Small Reasoning Models for Quantum Field Theory

Despite the growing application of Large Language Models (LLMs) to theoretical physics, there is little academic exploration into how domain-specific physics reasoning ability develops while training these models. To investigate this, we perform the first academic fine-tuning study of small (7B-parameter) reasoning models dedicated specifically to theoretical physics. Because open-source verifiable training data required to train such capabilities is scarce, we developed a robust data generation pipeline that can both create synthetic problems and make existing human-authored problems suitable for model training. Selecting Quantum Field Theory (QFT) as our primary domain, we generated over 2,500 synthetic problems alongside a curated collection of human-adapted problems sourced from arXiv and standard pedagogical resources. We conduct both Reinforcement Learning (RL) and Supervised Fine-Tuning (SFT) experiments, benchmarking performance gains as well as generalization to other physics domains. We perform an extensive analysis of model chains-of-though before and after fine-tuning, to understand how reasoning errors evolve during RL and SFT. Finally, we publicly release our data pipeline, verifiable QFT training data, and $\sim$200M tokens of QFT reasoning traces.

cs.LG↗

arXiv2026

Test-Time Scaling Makes Overtraining Compute-Optimal

Modern LLMs scale at test-time, e.g. via repeated sampling, where inference cost grows with model size and the number of samples. This creates a trade-off that pretraining scaling laws, such as Chinchilla, do not address. We present Train-to-Test ($T^2$) scaling laws that jointly optimize model size, training tokens, and number of inference samples under fixed end-to-end budgets. $T^2$ modernizes pretraining scaling laws with pass@$k$ modeling used for test-time scaling, then jointly optimizes pretraining and test-time decisions. Forecasts from $T^2$ are robust over distinct modeling approaches: measuring joint scaling effect on the task loss and modeling impact on task accuracy. Across eight downstream tasks, we find that when accounting for inference cost, optimal pretraining decisions shift radically into the overtraining regime, well-outside of the range of standard pretraining scaling suites. We validate our results by pretraining heavily overtrained models in the optimal region that $T^2$ scaling forecasts, confirming their substantially stronger performance compared to pretraining scaling alone. Finally, as frontier LLMs are post-trained, we show that our findings survive the post-training stage, making $T^2$ scaling meaningful in modern deployments.

cs.LG↗

arXiv2026

FAIR: Framing AIs Role in Programming Competitions -- Understanding How LLMs Are Changing the Game in Competitive Programming

This paper investigates how large language models (LLMs) are reshaping competitive programming. The field functions as an intellectual contest within computer science education and is marked by rapid iteration, real-time feedback, transparent solutions, and strict integrity norms. Prior work has evaluated LLMs performance on contest problems, but little is known about how human stakeholders -- contestants, problem setters, coaches, and platform stewards -- are adapting their workflows and contest norms under LLMs-induced shifts. At the same time, rising AI-assisted misuse and inconsistent governance expose urgent gaps in sustaining fairness and credibility. Drawing on 37 interviews spanning all four roles and a global survey of 207 contestants, as well as an API-based crawl of Codeforces contest logs (2022-2025) for quantitative analysis, we contribute: (i) an empirical account of evolving workflows, (ii) an analysis of contested fairness norms, and (iii) a chess-inspired governance approach with actionable measures -- real-time LLMs checks in online contests, peer co-monitoring and reporting, and cross-validation against offline performance -- to curb LLMs-assisted misuse while preserving fairness, transparency, and credibility.

cs.HC↗

arXiv2026

"It Talks Like a Patient, But Feels Different": Co-Designing AI Standardized Patients with Medical Learners

Standardized patients (SPs) play a central role in clinical communication training but are costly, difficult to scale, and inconsistent. Large language model (LLM) based AI standardized patients (AI-SPs) promise flexible, on-demand practice, yet learners often report that they talk like a patient but feel different. We interviewed 12 clinical-year medical students and conducted three co-design workshops to examine how learners experience constraints of SP encounters and what they expect from AI-SPs. We identified six learner-centered needs, translated them into AI-SP design requirements, and synthesized a conceptual workflow. Our findings position AI-SPs as tools for deliberate practice and show that instructional usability, rather than conversational realism alone, drives learner trust, engagement, and educational value.

cs.HC↗

arXiv2026

Prompting Destiny: Negotiating Socialization and Growth in an LLM-Mediated Speculative Gameworld

We present an LLM-mediated role-playing game that supports reflection on socialization, moral responsibility, and educational role positioning. Grounded in socialization theory, the game follows a four-season structure in which players guide a child prince through morally charged situations and compare the LLM-mediated NPC's differentiated responses across stages, helping them reason about how educational guidance shifts with socialization. To approximate real educational contexts and reduce score-chasing, the system hides real-time evaluative scores and provides delayed, end-of-stage growth feedback as reflective prompts. We conducted a user study (N=12) with gameplay logs and post-game interviews, analyzed via reflexive thematic analysis. Findings show how players negotiated responsibility and role positioning, and reveal an entry-load tension between open-ended expression and sustained engagement. We contribute design knowledge on translating sociological models of socialization into reflective AI-mediated game systems.

cs.HC↗

arXiv2026

SituFont: A Just-in-Time Adaptive Intervention System for Enhancing Mobile Readability in Situational Visual Impairments

Situational visual impairments (SVIs) hinder mobile readability, causing discomfort and limiting information access. Building on prior work in adaptive typography and accessibility, this paper presents SituFont, a context-aware and human-in-the-loop adaptive typography adjustment approach that enhances smartphone mobile readability by dynamically adjusting font parameters based on real-time contextual changes. Using smartphone sensors and a human-in-the-loop approach, SituFont personalizes text presentation to accommodate personal factors (e.g., fatigue, distraction) and environmental conditions (e.g., lighting, motion, location). To inform its design, we conducted formative interviews (N=15) to identify key SVI factors and controlled experiments (N=18) to quantify their impact on optimal text parameters. A comparative user study (N=12) across eight simulated SVI scenarios demonstrated SituFont's effectiveness in improving smartphone mobile readability in terms of improved efficiency and reduced workload compared with a non-trivial manual adjustment baseline.

cs.HC↗

arXiv2025

Evolution in Simulation: AI-Agent School with Dual Memory for High-Fidelity Educational Dynamics

Large language models (LLMs) based Agents are increasingly pivotal in simulating and understanding complex human systems and interactions. We propose the AI-Agent School (AAS) system, built around a self-evolving mechanism that leverages agents for simulating complex educational dynamics. Addressing the fragmented issues in teaching process modeling and the limitations of agents performance in simulating diverse educational participants, AAS constructs the Zero-Exp strategy, employs a continuous "experience-reflection-optimization" cycle, grounded in a dual memory base comprising experience and knowledge bases and incorporating short-term and long-term memory components. Through this mechanism, agents autonomously evolve via situated interactions within diverse simulated school scenarios. This evolution enables agents to more accurately model the nuanced, multi-faceted teacher-student engagements and underlying learning processes found in physical schools. Experiment confirms that AAS can effectively simulate intricate educational dynamics and is effective in fostering advanced agent cognitive abilities, providing a foundational stepping stone from the "Era of Experience" to the "Era of Simulation" by generating high-fidelity behavioral and interaction data.

cs.AI↗

arXiv2025

Pretrained Hybrids with MAD Skills

While Transformers underpin modern large language models (LMs), there is a growing list of alternative architectures with new capabilities, promises, and tradeoffs. This makes choosing the right LM architecture challenging. Recently proposed hybrid architectures seek a best-of-all-worlds approach that reaps the benefits of all architectures. Hybrid design is difficult for two reasons: it requires manual expert-driven search, and new hybrids must be trained from scratch. We propose Manticore, a framework that addresses these challenges by automating the design of hybrid architectures while reusing pretrained models to create pretrained hybrids. Our approach augments ideas from differentiable Neural Architecture Search (NAS) by incorporating simple projectors that translate features between pretrained blocks from different architectures. We then fine-tune hybrids that combine pretrained models from different architecture families -- such as the GPT series and Mamba -- end-to-end. With Manticore, we enable LM selection without training multiple models, the construction of pretrained hybrids from existing pretrained models, and the ability to program pretrained hybrids to have certain capabilities. Manticore hybrids match existing manually designed hybrids, achieve strong performance on Long Range Arena, and improve on pretrained transformers and state space models on various natural language tasks.

cs.LG↗

arXiv2025

PatternSight: A Perceptual Grouping Effectiveness Assessment Approach for Graphical Patterns in Charts

The boom in visualization generation tools has significantly lowered the threshold for chart authoring. Nevertheless, chart authors with an insufficient understanding of perceptual theories may encounter difficulties in evaluating the effectiveness of chart representations, thereby struggling to identify the appropriate chart design to convey the intended data patterns. To address this issue, we propose a perception simulation model that can assess the perceptual effectiveness of charts by predicting graphical patterns that chart viewers are likely to notice. The perception simulation model integrates perceptual theory into visual feature extraction of chart elements to provide interpretable model outcomes. Human perceptual results proved that the outcome of our model can simulate the perceptual grouping behaviors of most chart viewers and cover diverse perceptual results. We also embed the model into a prototype interface called PatternSight to facilitate chart authors in assessing whether the chart design can satisfy their pattern representation requirements as expected and determining feasible improvements of visual design. According to the results of a user experiment, PatternSight can effectively assist chart authors in optimizing chart design for representing data patterns.

cs.HC↗

arXiv2025

Test-time Scaling Techniques in Theoretical Physics -- A Comparison of Methods on the TPBench Dataset

Large language models (LLMs) have shown strong capabilities in complex reasoning, and test-time scaling techniques can enhance their performance with comparably low cost. Many of these methods have been developed and evaluated on mathematical reasoning benchmarks such as AIME. This paper investigates whether the lessons learned from these benchmarks generalize to the domain of advanced theoretical physics. We evaluate a range of common test-time scaling methods on the TPBench physics dataset and compare their effectiveness with results on AIME. To better leverage the structure of physics problems, we develop a novel, symbolic weak-verifier framework to improve parallel scaling results. Our empirical results demonstrate that this method significantly outperforms existing test-time scaling approaches on TPBench. We also evaluate our method on AIME, confirming its effectiveness in solving advanced mathematical problems. Our findings highlight the power of step-wise symbolic verification for tackling complex scientific problems.

cs.LG↗

arXiv2025

Characterizing Unintended Consequences in Human-GUI Agent Collaboration for Web Browsing

The proliferation of Large Language Model (LLM)-based Graphical User Interface (GUI) agents in web browsing scenarios present complex unintended consequences (UCs). This paper characterizes three UCs from three perspectives: phenomena, influence and mitigation, drawing on social media analysis (N=221 posts) and semi-structured interviews (N=14). Key phenomenon for UCs include agents' deficiencies in comprehending instructions and planning tasks, challenges in executing accurate GUI interactions and adapting to dynamic interfaces, the generation of unreliable or misaligned outputs, and shortcomings in error handling and feedback processing. These phenomena manifest as influences from unanticipated actions and user frustration, to privacy violations and security vulnerabilities, and further to eroded trust and wider ethical concerns. Our analysis also identifies user-initiated mitigation, such as technical adjustments and manual oversight, and provides implications for designing future LLM-based GUI agents that are robust, user-centric, and transparent, fostering a crucial balance between automation and human oversight.

cs.HC↗

arXiv2025

How to Make Your Multi-Image Posts Popular? An Approach to Enhanced Grid for Nine Images on Social Media

The nine-grid layout is commonly used for multi-image posts, arranging nine images in a tic-tac-toe board. This layout effectively presents content within limited space. Moreover, due to the numerous possible arrangements within the nine-image grid, the optimal arrangement that yields the highest level of attractiveness remains unknown. Our study investigates how the arrangement of images within a nine-grid layout affects the overall popularity of the image set, aiming to explore alignment schemes more aligned with user preferences. Based on survey results regarding user preferences in image arrangement, we have identified two ordering sequences that are widely recognized: sequential order and center prioritization, considering both image visual content and aesthetic quality as alignment metrics, resulting in four layout schemes. Finally, we recruited participants to annotate various layout schemes of the same set of images. Our experience-centered evaluation indicates that layout schemes based on aesthetic quality outperformed others. This research yields empirical evidence supporting the optimization of the nine-grid layout for multi-image posts, thereby furnishing content creators with valuable insights to enhance both attractiveness and user experience.

cs.HC↗

arXiv2025

Theoretical Physics Benchmark (TPBench) -- a Dataset and Study of AI Reasoning Capabilities in Theoretical Physics

We introduce a benchmark to evaluate the capability of AI to solve problems in theoretical physics, focusing on high-energy theory and cosmology. The first iteration of our benchmark consists of 57 problems of varying difficulty, from undergraduate to research level. These problems are novel in the sense that they do not come from public problem collections. We evaluate our data set on various open and closed language models, including o3-mini, o1, DeepSeek-R1, GPT-4o and versions of Llama and Qwen. While we find impressive progress in model performance with the most recent models, our research-level difficulty problems are mostly unsolved. We address challenges of auto-verifiability and grading, and discuss common failure modes. While currently state-of-the art models are still of limited use for researchers, our results show that AI assisted theoretical physics research may become possible in the near future. We discuss the main obstacles towards this goal and possible strategies to overcome them. The public problems and solutions, results for various models, and updates to the data set and score distribution, are available on the website of the dataset tpbench.org.

cs.LG↗

arXiv2024

Metamorpheus: Interactive, Affective, and Creative Dream Narration Through Metaphorical Visual Storytelling

Human emotions are essentially molded by lived experiences, from which we construct personalised meaning. The engagement in such meaning-making process has been practiced as an intervention in various psychotherapies to promote wellness. Nevertheless, to support recollecting and recounting lived experiences in everyday life remains under explored in HCI. It also remains unknown how technologies such as generative AI models can facilitate the meaning making process, and ultimately support affective mindfulness. In this paper we present Metamorpheus, an affective interface that engages users in a creative visual storytelling of emotional experiences during dreams. Metamorpheus arranges the storyline based on a dream's emotional arc, and provokes self-reflection through the creation of metaphorical images and text depictions. The system provides metaphor suggestions, and generates visual metaphors and text depictions using generative AI models, while users can apply generations to recolour and re-arrange the interface to be visually affective. Our experience-centred evaluation manifests that, by interacting with Metamorpheus, users can recall their dreams in vivid detail, through which they relive and reflect upon their experiences in a meaningful way.

cs.HC↗