Searcharxiv⌕ Search

arXiv subjects

Daoyuan Chen

Publications and source records attributed to Daoyuan Chen.

At least 19 recordsLinked to original sources
arXiv2026

A Unified Arena for LLM Agents across Diverse Eurogame

Interactive environments offer a practical way to evaluate large language model (LLM) agents, and competitive games extend this evaluation to decisions shaped by competing goals and other agents' choices. Yet existing evaluations confined to individual games or execution frameworks leave unclear which competitive advantages persist across settings. We introduce DECE (Diverse Eurogame Competitive Environments), a unified arena spanning ten Eurogames with shared interfaces and evaluation protocols. These games combine lasting strategic commitments, competition over shared resources, and multiple routes to victory. Our primary study evaluates 12 model-harness configurations, spanning nine models and four execution harnesses, in 480 three-player matches across ten games. The main and supplementary studies comprise 1,032 completed matches in total. These comparisons reveal broad cross-game ranking agreement alongside game-specific strengths. Matched studies show that the same model can perform differently across execution harnesses, adding another source of variation to these profiles. These performance advantages do not consistently translate into better token or cost efficiency, making resource use an additional consideration when comparing configurations. Together, the findings support evaluating complete agent configurations by both their competitive performance across games and the resources needed to achieve it.

cs.AI↗
arXiv2026

Rethinking Retrieval in the Long-Context Era: Training-Free Semantic Trees for Adaptive Context Construction

Long-context language models can increasingly process entire documents, making full-context prompting a practical alternative to passage retrieval. However, greater capacity determines how much information a model can receive, but not which information best supports a query. This distinction motivates rethinking retrieval: when the full document already fits, can selecting and organizing its evidence improve answering? The challenge is to retain supporting facts and relations while keeping the context focused on the question. To reconcile these goals, we introduce \ours, a training-free framework that uses semantic abstraction to guide selection while retaining original text for answering. \ours constructs a reusable semantic hierarchy from the full document in a single generation, linking node descriptions to source spans. Given a query, structural--semantic scoring and fine-grained lexical attribution nominate shared source nodes; their union with ancestor context preserves structural relations and complete local text. Across four frontier answer models, \ours improves average QA-F1 by 1.72 points over full-context prompting on three LongBench tasks, surpassing the evaluated baselines on the four-model average while reducing answer-context length by 58.55\%. Our code is available at the $\href{https://anonymous.4open.science/r/treefuse-ECFE}{repository}$.

cs.CL↗
arXiv2026

Agentic Data Engineering for LLMs: System Design and a Controlled Empirical Study

LLM data engineering requires iterative experimentation to determine how data selection, rewards, and curricula affect model learning, making it a natural setting for AI-for-AI (AI4AI) research. Can agents use experimental feedback to progressively improve strategies, and do the resulting strategies improve target-model performance on held-out data? We study how agents use experimental feedback to develop data-engineering strategies under fixed training and evaluation protocols. Our Agentic Data Engineering (ADE) framework lets agents pursue, revise, or combine research directions using cumulative findings. Strategies are selected by in-loop scores and evaluated on held-out benchmarks. Across four tasks-math and code data selection, math reward design, and math curriculum learning-selected strategies achieve absolute gains over generic baselines of up to 13.34% on AIME25 Pass@32 and 5.85% averaged across four held-out benchmarks under fixed training conditions. In-loop scores improve over successive experiments, while held-out trajectories show that later in-loop gains do not always improve generalization. Comparisons of research organization and feedback access favor combining evidence analysis with cumulative findings. Research trajectories show how earlier findings inform strategy revisions and combinations across research directions.

cs.CL↗
arXiv2026

Speculative Evaluation of Stochastic LLMs

Evaluating a stochastic large language model is costly: benchmark scores estimate expected performance from randomized rollouts, yet uniform repetition ignores sharp differences in task-level rollout variance. We ask how to minimize the variance of a fixed-benchmark mean under an exact rollout budget. We develop Speculative Evaluation with a Hierarchical Bayesian Neyman (HBN) policy with pilot size and stage weight jointly chosen ex ante. It runs a short uniform pilot, pools per-task success counts with a hierarchical Bayesian model, and uses posterior expectations of task-level sampling variances for exact positive-integer Neyman allocation. To mitigate the pilot synchronization barrier, HBN-async speculatively executes continuations from partial pilot feedback and retains those selected by the final allocation. Across six checkpoints and 18 benchmark groups, we evaluate 107 nondegenerate benchmark-checkpoint profiles. For rollout budgets of 8-64 per task, Speculative Evaluation reduces variance relative to Uniform by 12.8%-33.6% on average across profiles, outperforming hindsight-tuned empirical and independent Bayesian baselines. Real-generation experiments that account for the pilot synchronization barrier show that HBN-async mitigates its overhead, helping translate statistical efficiency into practical evaluation benefits.

stat.ML↗
arXiv2026

Attributing Structured-Output Gains in Function Calling: Interface Alignment versus Procedural Transfer

Structured-output benchmarks reward both task decisions and interface compliance, so prompt-induced function-calling gains require attribution before they can be interpreted as transferable skill. We introduce a four-layer gain-attribution protocol for prompt-prepended skill injection, combining canonicalized rescoring, format-only controls, repaired/balanced induction, and portability checks. Applied to the Berkeley Function Calling Leaderboard (BFCL) and scoped with API-Bank, MATH-500, and MultiHop-RAG, the protocol shows that several apparent gains are better attributed to interface alignment than to procedural transfer: format-only prompts match or exceed full skills in key BFCL cells, repaired/balanced induction removes the largest sub-frontier gains, and API-Bank target-native gains are matched within 0.5 percentage points (pp) by length-matched generic procedural prompts. These findings treat format compliance as a useful engineering capability while clarifying what a structured-output score certifies. We release BFCL-CANONICAL and recommend canonicalized metrics, balanced induction, and format-only baselines for function-calling skill-gain attribution. Code and data are available at https://github.com/couragec/skill-injection-attribution.

cs.SE↗
arXiv2026

Beyond Solvability: Task Learnability as a Static Prior for LLM RL Post-Training

Reinforcement learning (RL) has become a central post-training paradigm for eliciting reasoning capabilities in large language models, yet uniform task sampling allocates compute without regard to differences in how tasks respond to optimization. Existing task-valuation methods mostly rely on snapshot-based signals such as current pass rate or reward, which estimate how solvable a task is under the current policy. However, tasks with similar current solvability can still differ substantially in how positively they respond to further training. We study this residual axis as task learnability: a regime-conditional measure of expected positive response to continued training under a fixed RL post-training regime. By analyzing per-task reward trajectories, we find that learnability is reproducible across independently sampled training contexts and predictive of downstream utility. To make this signal practical before training begins, we propose TrajVal, a lightweight probe-based estimator that approximates per-task learnability from a short probe run and two endpoint evaluations. TrajVal can be used either as a standalone static prior for task sampling or as a multiplicative prior for existing online schedulers. Experiments on mathematical and logical reasoning benchmarks across multiple model scales show that TrajVal improves data efficiency over uniform sampling and provides complementary gains when combined with online scheduling methods.

cs.LG↗
arXiv2026

CDR-Bench: Evaluating Faithful Execution of Compositional, Order-Sensitive Data Refinement Recipes

Data refinement involves executing multi-step recipes over evolving text states, where both composition and execution order of processing operators determine the outcome. While existing benchmarks either isolate text editing or entangle it with code and tool execution, it remains unclear whether LLMs can directly and faithfully execute these compositional, order-sensitive data refinement recipes. To fill this gap, we introduce CDR-Bench, a comprehensive benchmark featuring 3,462 high-quality tasks spanning four real-world data refinement domains and 29 distinct operators. Our benchmark evaluates models across atomic, order-agnostic, and order-sensitive settings, leveraging deterministic reference outputs to enable exact evaluation. Experiments on 10+ state-of-the-art LLMs reveal consistent failure patterns: performance degrades sharply in compositional settings, and order-sensitive recipe success collapses. These findings underline that current LLMs lack the procedural faithfulness required for reliable compositional data refinement.

cs.AI↗
arXiv2026

GEOALIGN: Geometric Rollout Curation for Robust LLM Reinforcement Learning

Online reinforcement learning is widely used to align large language models (LLMs) with reward signals, yet training can be unstable under noisy or misspecified rewards. We identify a failure mode we call directional inconsistency: within a batch, a small set of high-reward rollouts induces representation-space preference directions that sharply disagree with the batch majority, resulting in high-variance and destabilizing updates. We propose geoalign, a lightweight plug-in for rollout curation in iterative policy optimization. Geoalign (i) forms within-prompt preference pairs, (ii) learns an online projector on per-rollout hidden states to concentrate reward-ordered displacement directions, and (iii) detects directionally inconsistent rollouts via their angular deviation from a batch consensus prototype and rectifies them with within-prompt stable alternatives. Geoalign is forward-pass only and adds negligible overhead. Across dialogue alignment with a learned reward model and mathematical reasoning with binary verified rewards, Geoalign improves final performance and reduces training oscillation, outperforming PF-PPO, PAR, PODS, and Seed-GRPO. These results suggest latent directional consensus as an effective reliability signal for online LLM RL.

cs.LG↗
arXiv2026

DetailMaster: Can Your Text-to-Image Model Handle Long Prompts?

While recent Text-to-Image (T2I) models show impressive capabilities in synthesizing images from brief descriptions, they struggle with the long, detailed prompts required for professional applications. We present DetailMaster, a comprehensive benchmark for evaluating T2I capabilities on long prompts with complex compositional requirements, accompanied by an automated data construction pipeline and an evaluation workflow. Comprising expert-validated prompts averaging 284.89 tokens, our benchmark introduces four critical evaluation dimensions: Character Attributes, Structured Character Locations, Multi-Dimensional Scene Attributes, and Spatial/Interactive Relationships. Evaluations on various general-purpose and long-prompt-optimized models reveal critical performance limitations, showing that weak encoders struggle to preserve syntactic dependencies within prompts and diffusion models suffer from attribute leakage under detail-intensive conditions. Through a controlled ablation study under varying constraints, we further show that high-fidelity generation requires a synergistic combination of expanded prompt limits and long-prompt training. We open-source our dataset and code to foster progress in long-prompt-driven T2I generation.

cs.CV↗
arXiv2026

HumanVBench: Probing Human-Centric Video Understanding in MLLMs with Automatically Synthesized Benchmarks

Evaluating the nuanced human-centric video understanding capabilities of Multimodal Large Language Models (MLLMs) remains a great challenge, as existing benchmarks often overlook the intricacies of emotion, behavior, and cross-modal alignment. We introduce HumanVBench, a comprehensive video benchmark designed to rigorously probe these capabilities across 16 fine-grained tasks. A cornerstone of our work is a novel and scalable benchmark construction methodology, featuring two automated pipelines that synthesize high-quality video annotations and challenging multiple-choice questions with minimal human labor. By leveraging state-of-the-art models for annotation and systematically converting model-induced errors into plausible distractors, our framework provides a generalizable ``machine'' for creating nuanced evaluation suites. Our extensive evaluation of 30 leading MLLMs on HumanVBench reveals critical deficiencies, particularly in perceiving subtle emotions and aligning speech with visual cues, with even top proprietary models falling short of human performance. We open-source HumanVBench and our synthesis pipelines to catalyze the development of more socially intelligent and capable video MLLMs.

cs.CV↗
arXiv2026

VeriSciQA: An Auto-Verified Dataset for Scientific Visual Question Answering

Large Vision-Language Models (LVLMs) show promise for scientific applications, yet open-source models still struggle with Scientific Visual Question Answering (SVQA), namely answering questions about figures from scientific papers. A key bottleneck is the lack of public, large-scale, high-quality SVQA datasets. Although recent work uses LVLMs to synthesize data at scale, we identify systematic errors in their resulting QA pairs, stemming from LVLMs' inherent limitations and information asymmetry between figures and text. To address these challenges, we propose a Cross-Modal verification framework that generates questions and answers purely from figure-citing paragraphs, then verifies them against the figures themselves, leveraging the inherent text-figure alignment in scientific papers to filter out erroneous QA pairs. We instantiate this framework to curate VeriSciQA, a dataset of 20,272 QA pairs spanning 20 scientific domains and 12 figure types. Difficulty assessment reveals a notable accuracy gap between the best open-source model (65%) and the best proprietary model (80.5%), demonstrating room for improvement. Moreover, models fine-tuned on VeriSciQA achieve consistent improvements on SVQA benchmarks, with performance gains that scale with data size, surpassing models trained on existing datasets. Human evaluation further validates the improved quality of VeriSciQA. These results demonstrate that continued data expansion via our scalable framework can further advance SVQA capability in the open-source community. Our dataset is publicly available at https://huggingface.co/datasets/datajuicer/VeriSciQA.

cs.CV↗
arXiv2026

Spectral Evolution Search: Efficient Inference-Time Scaling for Reward-Aligned Image Generation

Inference-time scaling offers a versatile paradigm for aligning visual generative models with downstream objectives without parameter updates. However, existing approaches that optimize the high-dimensional initial noise suffer from severe inefficiency, as many search directions exert negligible influence on the final generation. We show that this inefficiency is closely related to a spectral bias in generative dynamics: model sensitivity to initial perturbations diminishes rapidly as frequency increases. Building on this insight, we propose Spectral Evolution Search (SES), a plug-and-play framework for initial noise optimization that executes gradient-free evolutionary search within a low-frequency subspace. Theoretically, we derive the Spectral Scaling Prediction from perturbation propagation dynamics, which explains the systematic differences in the impact of perturbations across frequencies. Extensive experiments demonstrate that SES significantly advances the Pareto frontier of generation quality versus computational cost, consistently outperforming strong baselines under equivalent budgets.

cs.LG↗
arXiv2026

VIRAL: Visual In-Context Reasoning via Analogy in Diffusion Transformers

Replicating In-Context Learning (ICL) in computer vision remains challenging due to task heterogeneity. We propose \textbf{VIRAL}, a framework that elicits visual reasoning from a pre-trained image editing model by formulating ICL as conditional generation via visual analogy ($x_s : x_t :: x_q : y_q$). We adapt a frozen Diffusion Transformer (DiT) using role-aware multi-image conditioning and introduce a Mixture-of-Experts LoRA to mitigate gradient interference across diverse tasks. Additionally, to bridge the gaps in current visual context datasets, we curate a large-scale dataset spanning perception, restoration, and editing. Experiments demonstrate that VIRAL outperforms existing methods, validating that a unified V-ICL paradigm can handle the majority of visual tasks, including open-domain editing. Our code is available at https://anonymous.4open.science/r/VIRAL-744A

cs.CV↗
arXiv2026

BOTS: A Unified Framework for Bayesian Online Task Selection in LLM Reinforcement Finetuning

Reinforcement finetuning (RFT) is a key technique for aligning Large Language Models (LLMs) with human preferences and enhancing reasoning, yet its effectiveness is highly sensitive to which tasks are explored during training. Uniform task sampling is inefficient, wasting computation on tasks that are either trivial or unsolvable, while existing task selection methods often suffer from high rollout costs, poor adaptivity, or incomplete evidence. We introduce BOTS, a unified framework for Bayesian Online Task Selection in LLM reinforcement finetuning. Grounded in Bayesian inference, BOTS adaptively maintains posterior estimates of task difficulty as the model evolves. It jointly incorporates explicit evidence from direct evaluations of selected tasks and implicit evidence inferred from these evaluations for unselected tasks, with Thompson sampling ensuring a principled balance between exploration and exploitation for task selection. To make implicit evidence practical, we instantiate it with an ultra-light interpolation-based plug-in that estimates difficulties of tasks without extra rollouts, adding negligible overhead. Empirically, across diverse domains and LLM scales, BOTS consistently improves data efficiency and performance over baselines and ablations, providing a practical and extensible solution for dynamic task selection in RFT. Code is available at https://github.com/agentscope-ai/Trinity-RFT/tree/main/examples/bots.

cs.AI↗
arXiv2026

AttriCtrl: Fine-Grained Control of Aesthetic Attribute Intensity in Diffusion Models

Diffusion models have recently become the dominant paradigm for image generation, yet existing systems struggle to interpret and follow numeric instructions for adjusting semantic attributes. In real-world creative scenarios, especially when precise control over aesthetic attributes is required, current methods fail to provide such controllability. This limitation partly arises from the subjective and context-dependent nature of aesthetic judgments, but more fundamentally stems from the fact that current text encoders are designed for discrete tokens rather than continuous values. Meanwhile, efforts on aesthetic alignment, often leveraging reinforcement learning, direct preference optimization, or architectural modifications, primarily align models with a global notion of human preference. While these approaches improve user experience, they overlook the multifaceted and compositional nature of aesthetics, underscoring the need for explicit disentanglement and independent control of aesthetic attributes. To address this gap, we introduce AttriCtrl, a lightweight framework for continuous aesthetic intensity control in diffusion models. It first defines relevant aesthetic attributes, then quantifies them through a hybrid strategy that maps both concrete and abstract dimensions onto a unified $[0,1]$ scale. A plug-and-play value encoder is then used to transform user-specified values into model-interpretable embeddings for controllable generation. Experiments show that AttriCtrl achieves accurate and continuous control over both single and multiple aesthetic attributes, significantly enhancing personalization and diversity. Crucially, it is implemented as a lightweight adapter while keeping the diffusion model frozen, ensuring seamless integration with existing frameworks such as ControlNet at negligible computational cost.

cs.CV↗
arXiv2025

FlowerTune: A Cross-Domain Benchmark for Federated Fine-Tuning of Large Language Models

Large Language Models (LLMs) have achieved state-of-the-art results across diverse domains, yet their development remains reliant on vast amounts of publicly available data, raising concerns about data scarcity and the lack of access to domain-specific, sensitive information. Federated Learning (FL) presents a compelling framework to address these challenges by enabling decentralized fine-tuning on pre-trained LLMs without sharing raw data. However, the compatibility and performance of pre-trained LLMs in FL settings remain largely under explored. We introduce the FlowerTune LLM Leaderboard, a first-of-its-kind benchmarking suite designed to evaluate federated fine-tuning of LLMs across four diverse domains: general NLP, finance, medical, and coding. Each domain includes federated instruction-tuning datasets and domain-specific evaluation metrics. Our results, obtained through a collaborative, open-source and community-driven approach, provide the first comprehensive comparison across 26 pre-trained LLMs with different aggregation and fine-tuning strategies under federated settings, offering actionable insights into model performance, resource constraints, and domain adaptation. This work lays the foundation for developing privacy-preserving, domain-specialized LLMs for real-world applications.

cs.CL↗
arXiv2025

Grounded in Reality: Learning and Deploying Proactive LLM from Offline Logs

Large Language Models (LLMs) excel as passive responders, but teaching them to be proactive, goal-oriented partners, a critical capability in high-stakes domains, remains a major challenge. Current paradigms either myopically optimize single-turn attributes or rely on brittle, high-cost user simulators, creating a persistent ``reality gap''. To bridge this gap, we introduce \texttt{Learn-to-Ask}, a general, simulator-free framework for learning and deploying proactive dialogue agents \textit{directly from offline expert data}, bypassing the need to model complex user dynamics. Our key insight is to reframe the offline policy learning problem by leveraging the \textbf{observed future} of each expert trajectory. This allows us to infer a dense, turn-by-turn reward signal grounded in the expert's revealed strategy, decomposing the intractable long-horizon problem into a series of supervised learning tasks, and training a policy to output a structured \texttt{(action, state_assessment)} tuple, governing both \textbf{what to ask} and, crucially, \textbf{when to stop}. To ensure reward fidelity, our Automated Grader Calibration pipeline systematically purges noise from the LLM-based reward model with minimal human supervision. Empirically, we demonstrate the efficacy of \texttt{Learn-to-Ask} in a real-world medical dataset, using LLMs of varying sizes up to 32B. Our approach culminates in the successful deployment of LLMs into a live, large-scale online AI service. In rigorous in-house evaluations, our model was launched and achieved performance even superior to human experts, proving our framework's ability to translate offline data into tangible, real-world impact. We hope this work provides a practical and economically viable blueprint for transforming passive LLMs into proactive, goal-oriented LLM applications.

cs.CL↗
arXiv2025

Diversity as a Reward: Fine-Tuning LLMs on a Mixture of Domain-Undetermined Data

Fine-tuning large language models (LLMs) using diverse datasets is crucial for enhancing their overall performance across various domains. In practical scenarios, existing methods based on modeling the mixture proportions of data composition often struggle with data whose domain labels are missing, imprecise or non-normalized, while methods based on data selection usually encounter difficulties in balancing multi-domain performance. To address these challenges, in this work, we investigate the role of data diversity in enhancing the overall abilities of LLMs by empirically constructing contrastive data pools and theoretically deriving explanations. Building upon the insights gained, we propose a new method that gives the LLM a dual identity: an output model to cognitively probe and select data based on diversity reward, as well as an input model to be tuned with the selected data. Extensive experiments show that the proposed method notably boosts performance across domain-undetermined data and a series of foundational downstream tasks when applied to various advanced LLMs. We release our code and hope this study can shed light on the understanding of data diversity and advance feedback-driven data-model co-design for LLMs.

cs.CL↗