paper-with-me

홈 › Papers

Discovering Knowledge Deficiencies of Language Models on Massive Knowledge Base

2025-03-30 · Linxin Song, Xuwei Ding, Jieyu Zhang, Taiwei Shi, Ryotaro Shimizu, Rahul Gupta, Yang Liu, Jian Kang, Jieyu Zhao

Large language models (LLMs) possess impressive linguistic capabilities but often fail to faithfully retain factual knowledge, leading to hallucinations and unreliable outputs. Understanding LLMs' knowledge deficiencies by exhaustively evaluating against full-scale knowledge bases is computationally prohibitive, especially for closed-weight models. We propose stochastic error ascent (SEA), a scalable and efficient framework for discovering knowledge deficiencies (errors) in closed-weight LLMs under a strict query budget. Rather than naively probing all knowledge candidates, SEA formulates error discovery as a stochastic optimization process: it iteratively retrieves new high-error candidates by leveraging the semantic similarity to previously observed failures. To further enhance search efficiency and coverage, SEA employs hierarchical retrieval across document and paragraph levels, and constructs a relation directed acyclic graph to model error propagation and identify systematic failure modes. Empirically, SEA uncovers 40.7x more knowledge errors than Automated Capability Discovery and 26.7% more than AutoBencher, while reducing the cost-per-error by 599x and 9x, respectively. Human evaluation confirms the high quality of generated questions, while ablation and convergence analyses validate the contribution of each component in SEA. Further analysis on the discovered errors reveals correlated failure patterns across LLM families and recurring deficits, highlighting the need for better data coverage and targeted fine-tuning in future LLM development.

📄 PDF Abstract BibTeX arXiv:2503.23361

Code (0)

등록된 구현이 없습니다.

Tasks

Semantic SimilaritySemantic Textual SimilarityStochastic Optimization

Similar Papers 제목 키워드 기반

Diagnosing and Remedying Knowledge Deficiencies in LLMs via Label-free Curricular Meaningful Learning

2024-08-21 · Kai Xiong, Xiao Ding, Li Du, Jiahao Ying 외

Large Language Models (LLMs) are versatile and demonstrate impressive generalization ability by mining and learning information from extensive unlabeled text. However, they still exhibit reasoning mistakes, often stemmin…

Diagnostic

Structural Entropy Guided Agent for Detecting and Repairing Knowledge Deficiencies in LLMs

2025-05-12 · Yifan Wei, Xiaoyan Yu, Tengfei Pan, Angsheng Li 외

Large language models (LLMs) have achieved unprecedented performance by leveraging vast pretraining corpora, yet their performance remains suboptimal in knowledge-intensive domains such as medicine and scientific researc…

AI AgentKnowledge DistillationKnowledge GraphsReinforcement Learning+1

Enriching Basque Coreference Resolution System using Semantic Knowledge sources

2017-04-01 · WS 2017 4 · Ander Soraluze, Olatz Arregi, Xabier Arregi, Arantza D{\'\i}az de Ilarraza

In this paper we present a Basque coreference resolution system enriched with semantic knowledge. An error analysis carried out revealed the deficiencies that the system had in resolving coreference cases in which semant…

coreference-resolutionCoreference ResolutionWorld Knowledge

Deep Expert Injection for Anchoring Retinal VLMs with Domain-Specific Knowledge

2026-03-07 · Shuai Lu, Meng Wang, Jia Guo, Jiawei Du 외 arxiv

Large Vision Language Models (LVLMs) show immense potential for automated ophthalmic diagnosis. However, their clinical deployment is severely hindered by lacking domain-specific knowledge. In this work, we identify two …

Visual Question Answering

Untying the Reversal Curse via Bidirectional Language Model Editing

2023-10-16 · Jun-Yu Ma, Jia-Chen Gu, Zhen-Hua Ling, Quan Liu 외

Recent studies have demonstrated that large language models (LLMs) store massive factual knowledge within their parameters. But existing LLMs are prone to hallucinate unintended text due to false or outdated knowledge. S…

knowledge editingLanguage ModelingLanguage ModellingModel Editing+1