arXiv subjects

Zhiyu Shen

Publications and source records attributed to Zhiyu Shen.

5 recordsLinked to original sources

arXiv2026

Rethinking Reverse KL as Adaptive Entropy Distillation

Knowledge distillation (KD) is widely used to transfer the capabilities of large language models (LLMs) to smaller students, but existing objectives often struggle to balance faithful imitation and robust generation. In particular, existing methods mainly combine FKL and RKL, overlooking that RKL itself provides a mechanism for adjusting the student's imitation strength. Motivated by this, we revisit on-policy Reverse Kullback-Leibler (RKL) distillation and decompose its objective into a teacher-fitting term and a student-entropy term, without introducing an explicit FKL branch. We show theoretically that the token-level optimal student distribution corresponds to a tempered variant of the teacher distribution, where the adaptive weight controls the trade-off between mode-seeking and uncertainty preservation. Guided by this insight, we propose \textbf{Adaptive Entropy Distillation (AED)}, which uses the teacher's entropy to dynamically calibrate token-level imitation strength. Experiments on instruction-following and mathematical reasoning benchmarks demonstrate that AED achieves superior overall performance and generally improves teacher--student distributional and entropy alignment.

cs.LG↗

arXiv2026

MemBuilder: Reinforcing LLMs for Long-Term Memory Construction via Attributed Dense Rewards

Maintaining consistency in long-term dialogues remains a fundamental challenge for LLMs, as standard retrieval mechanisms often fail to capture the temporal evolution of historical states. While memory-augmented frameworks offer a structured alternative, current systems rely on static prompting of closed-source models or suffer from ineffective training paradigms with sparse rewards. We introduce MemBuilder, a reinforcement learning framework that trains models to orchestrate multi-dimensional memory construction with attributed dense rewards. MemBuilder addresses two key challenges: (1) Sparse Trajectory-Level Rewards: we employ synthetic session-level question generation to provide dense intermediate rewards across extended trajectories; and (2) Multi-Dimensional Memory Attribution: we introduce contribution-aware gradient weighting that scales policy updates based on each component's downstream impact. Experimental results show that MemBuilder enables a 4B-parameter model to outperform state-of-the-art closed-source baselines, exhibiting strong generalization across long-term dialogue benchmarks.

cs.CL↗

arXiv2026

Cross-Source Reasoning-based Correction for Author Name Disambiguation

Author name disambiguation is a critical challenge in academic search systems, often addressed through from-scratch and real-time disambiguation approaches. However, current algorithms remain vulnerable to cumulative errors of paper-author assignments and overlook inconsistent assignments across different sources. Resorting to expert annotation is resource-intensive. To this end, this paper explores a new perspective for author name disambiguation: cross-source correction by leveraging inconsistent assignments across sources. We propose CrossND, a full-stack framework that integrates data refinement, cross-source reasoning, and test-time scaling. First, a chain-of-refinement pipeline denoises author profiles and produces more accurate paper-author matching probabilities. Second, a supervised fine-tuning process incorporates these refined signals and a probabilistic soft logic-based cross-correction module to infer the assignments of which sources are incorrect. Third, test-time scaling further enhances the accuracy and robustness of the predictions. Experiments on real-world datasets indicate that CrossND consistently outperforms 17 baselines by leveraging cross-source reasoning without human intervention.

cs.CL↗

arXiv2026

HopWeaver: Cross-Document Synthesis of High-Quality and Authentic Multi-Hop Questions

Multi-Hop Question Answering (MHQA) is crucial for evaluating the model's capability to integrate information from diverse sources. However, creating extensive and high-quality MHQA datasets is challenging: (i) manual annotation is expensive, and (ii) current synthesis methods often produce simplistic questions or require extensive manual guidance. This paper introduces HopWeaver, the first cross-document framework synthesizing authentic multi-hop questions without human intervention. HopWeaver synthesizes bridge and comparison questions through an innovative pipeline that identifies complementary documents and constructs authentic reasoning paths to ensure true multi-hop reasoning. We further present a comprehensive system for evaluating the synthesized multi-hop questions. Empirical evaluations demonstrate that the synthesized questions achieve comparable or superior quality to human-annotated datasets at a lower cost. Our framework provides a valuable tool for the research community: it can automatically generate challenging benchmarks from any raw corpus, which opens new avenues for both evaluation and targeted training to improve the reasoning capabilities of advanced question answering models, especially in domains with scarce resources.

cs.CL↗

arXiv2023

DesignQuizzer: A Community-Powered Conversational Agent for Learning Visual Design

Online design communities, where members exchange free-form views on others' designs, offer a space for beginners to learn visual design. However, the content of these communities is often unorganized for learners, containing many redundancies and irrelevant comments. In this paper, we propose a computational approach for leveraging online design communities to run a conversational agent that assists informal learning of visual elements (e.g., color and space). Our method extracts critiques, suggestions, and rationales on visual elements from comments. We present DesignQuizzer, which asks questions about visual design in UI examples and provides structured comment summaries. Two user studies demonstrate the engagement and usefulness of DesignQuizzer compared with the baseline (reading reddit.com/r/UI_design). We also showcase how effectively novices can apply what they learn with DesignQuizzer in a design critique task and a visual design task. We discuss how to use our approach with other communities and offer design considerations for community-powered learning support tools.

cs.HC↗