arXiv subjects

Jianhuan Zhuo

Publications and source records attributed to Jianhuan Zhuo.

6 recordsLinked to original sources

arXiv2026

SWE-Journey: Towards More Realistic Evaluation of Coding Assistants through Long-Horizon, Multi-Turn Interaction

Coding assistants such as Claude Code and Codex have become a major application of LLM agents, yet existing benchmarks remain far from real-world use, particularly in task horizon and interaction length. Code assistants require completing long chains of development work in continuously evolving repositories, while repeatedly clarifying requirements and adapting implementations through multi-turn interaction. To address these gaps, we introduce SWE-Journey, a benchmark for more realistic evaluation of coding assistants. To address the task-horizon gap, we propose a weak-to-strong synthesis pipeline that automatically constructs long-horizon coding tasks. To address the interaction gap, we mine four representative user personas from real interaction data and build a user-simulation agent to reproduce realistic code-assistance interactions. On average, models pass over 75% of tests for requested functionality with software architects, but fewer than 25% with non-coders. These results show that current coding assistants still fall short of enabling reliable coding for non-coders. We further analyze the reasons for this gap and identify asking right, finding right, and fixing right as key capabilities during interaction.

cs.CL↗

arXiv2026

AutoGUIWorld: Image Generators as Visual World Models for GUI Agent

GUI agents require high-quality interaction trajectories to learn how software environments respond to actions, maintain state, and support multi-step workflows. However, the diversity of available trajectories is constrained by the applications, interface states, and workflows accessible in the underlying environments. Expanding this coverage requires deploying increasingly diverse and complex software, with specialized applications imposing additional installation, configuration, and runtime costs. We introduce AutoGUIWorld, a data generation framework that combines the visual priors of image generators with the task knowledge of a planner to synthesize GUI interaction trajectories without deploying or running the corresponding software environments. AutoGUIWorld samples initial GUI scenes from structured specifications of operating-system context, visual appearance, and interface state, and generates tasks conditioned on those scenes. A planner then specifies atomic actions and their intended visual consequences, while an image generator iteratively edits the current screenshot to produce subsequent observations. Action grounding and transition-level quality filtering yield 79,266 spatially annotated step-level training samples across Ubuntu, Windows, macOS, and Chrome. Fine-tuning Qwen3.5-35B-A3B on AutoGUIWorld trajectories improves the mean task score on OSWorld from 33.0% to 40.8% and the task success rate on ScienceBoard from 14.0% to 32.2%. These results show that generated trajectories improve GUI-agent performance on real desktop and scientific tasks.

cs.CV↗

arXiv2026

GRAPE: Let GRPO Supervise Query Rewriting by Ranking for Retrieval

The CLIP model has established itself as a cornerstone of large-scale retrieval systems. However, its performance often degrades under distributional shifts such as multilingual, long-form, or multimodal queries. To avoid the prohibitive costs associated with retriever retraining or corpus re-embedding, we propose GRAPE (Grouped Ranking-Aware Policy Optimization Enhancement), a plug-and-play approach that leverages LLM-based query rewriting to bridge these gaps. Unlike existing methods that lack explicit supervision, GRAPE integrates ranking signals into the rewriting LLM via Grouped Relative Policy Optimization (GRPO), ensuring rewritten queries are better aligned with the frozen retriever's latent distribution. Crucially, we identify a score inflation phenomenon in naive similarity-based finetuning - where irrelevant candidates receive indiscriminately high scores - and mitigate it with a novel corpus-relative ranking-based reward. Extensive experiments across multilingual (Flickr30k-CN, CVLUE, XM3600), long-form (Wikipedia), and multimodal (CIRR) benchmarks demonstrate that GRAPE consistently improves performance, achieving an average gain of 4.9% in Recall@10 without any modification to the underlying retriever. The code is available at https://github.com/mogulzhang/GRAPE.

cs.CV↗

arXiv2026

FBCIR: Balancing Cross-Modal Focuses in Composed Image Retrieval

Composed image retrieval (CIR) requires multi-modal models to jointly reason over visual content and semantic modifications presented in text-image input pairs. While current CIR models achieve strong performance on common benchmark cases, their accuracies often degrades in more challenging scenarios where negative candidates are semantically aligned with the query image or text. In this paper, we attribute this degradation to focus imbalances, where models disproportionately attend to one modality while neglecting the other. To validate this claim, we propose FBCIR, a multi-modal focus interpretation method that identifies the most crucial visual and textual input components to a model's retrieval decisions. Using FBCIR, we report that focus imbalances are prevalent in existing CIR models, especially under hard negative settings. Building on the analyses, we further propose a CIR data augmentation workflow that facilitates existing CIR datasets with curated hard negatives designed to encourage balanced cross-modal reasoning. Extensive experiments across multiple CIR models demonstrate that the proposed augmentation consistently improves performance in challenging cases, while maintaining their capabilities on standard benchmarks. Together, our interpretation method and data augmentation workflow provide a new perspective on CIR model diagnosis and robustness improvements.

cs.CV↗

arXiv2025

FailureAtlas:Mapping the Failure Landscape of T2I Models via Active Exploration

Static benchmarks have provided a valuable foundation for comparing Text-to-Image (T2I) models. However, their passive design offers limited diagnostic power, struggling to uncover the full landscape of systematic failures or isolate their root causes. We argue for a complementary paradigm: active exploration. We introduce FailureAtlas, the first framework designed to autonomously explore and map the vast failure landscape of T2I models at scale. FailureAtlas frames error discovery as a structured search for minimal, failure-inducing concepts. While it is a computationally explosive problem, we make it tractable with novel acceleration techniques. When applied to Stable Diffusion models, our method uncovers hundreds of thousands of previously unknown error slices (over 247,000 in SD1.5 alone) and provides the first large-scale evidence linking these failures to data scarcity in the training set. By providing a principled and scalable engine for deep model auditing, FailureAtlas establishes a new, diagnostic-first methodology to guide the development of more robust generative AI. The code is available at https://github.com/cure-lab/FailureAtlas

cs.CV↗

arXiv2021

Learning Explicit User Interest Boundary for Recommendation

The core objective of modelling recommender systems from implicit feedback is to maximize the positive sample score $s_p$ and minimize the negative sample score $s_n$, which can usually be summarized into two paradigms: the pointwise and the pairwise. The pointwise approaches fit each sample with its label individually, which is flexible in weighting and sampling on instance-level but ignores the inherent ranking property. By qualitatively minimizing the relative score $s_n - s_p$, the pairwise approaches capture the ranking of samples naturally but suffer from training efficiency. Additionally, both approaches are hard to explicitly provide a personalized decision boundary to determine if users are interested in items unseen. To address those issues, we innovatively introduce an auxiliary score $b_u$ for each user to represent the User Interest Boundary(UIB) and individually penalize samples that cross the boundary with pairwise paradigms, i.e., the positive samples whose score is lower than $b_u$ and the negative samples whose score is higher than $b_u$. In this way, our approach successfully achieves a hybrid loss of the pointwise and the pairwise to combine the advantages of both. Analytically, we show that our approach can provide a personalized decision boundary and significantly improve the training efficiency without any special sampling strategy. Extensive results show that our approach achieves significant improvements on not only the classical pointwise or pairwise models but also state-of-the-art models with complex loss function and complicated feature encoding.

cs.IR↗