paper-with-me

Papers

Probing Language Models on Their Knowledge Source

2024-10-08 · Zineddine Tighidet, Andrea Mogini, Jiali Mei, Benjamin Piwowarski, Patrick Gallinari

Large Language Models (LLMs) often encounter conflicts between their learned, internal (parametric knowledge, PK) and external knowledge provided during inference (contextual knowledge, CK). Understanding how LLMs models prioritize one knowledge source over the other remains a challenge. In this paper, we propose a novel probing framework to explore the mechanisms governing the selection between PK and CK in LLMs. Using controlled prompts designed to contradict the model's PK, we demonstrate that specific model activations are indicative of the knowledge source employed. We evaluate this framework on various LLMs of different sizes and demonstrate that mid-layer activations, particularly those related to relations in the input, are crucial in predicting knowledge source selection, paving the way for more reliable models capable of handling knowledge conflicts effectively.

📄 PDF Abstract BibTeX arXiv:2410.05817

Code (1)

Zineddine-Tighidet/knowledge-probing-framework 공식 구현

Tasks

Probing Language Models

Similar Papers 제목 키워드 기반

How to Probe Sentence Embeddings in Low-Resource Languages: On Structural Design Choices for Probing Task Evaluation

2020-06-16 · CONLL 2020 · Steffen Eger, Johannes Daxenberger, Iryna Gurevych

Sentence encoders map sentences to real valued vectors for use in downstream applications. To peek into these representations - e.g., to increase interpretability of their results - probing tasks have been designed which…

SentenceSentence Embeddings

Rewire-then-Probe: A Contrastive Recipe for Probing Biomedical Knowledge of Pre-trained Language Models

2021-11-16 · ACL ARR November 2021 11 · Anonymous

Knowledge probing is crucial for understanding the knowledge transfer mechanism behind the pre-trained language models (PLMs). Despite the growing progress of probing knowledge for PLMs in the general domain, specialised…

Knowledge ProbingTransfer Learning

Rewire-then-Probe: A Contrastive Recipe for Probing Biomedical Knowledge of Pre-trained Language Models

2021-10-15 · ACL 2022 5 · Zaiqiao Meng, Fangyu Liu, Ehsan Shareghi, Yixuan Su 외

Knowledge probing is crucial for understanding the knowledge transfer mechanism behind the pre-trained language models (PLMs). Despite the growing progress of probing knowledge for PLMs in the general domain, specialised…

Knowledge ProbingTransfer Learning

Give Me the Facts! A Survey on Factual Knowledge Probing in Pre-trained Language Models

2023-10-25 · Paul Youssef, Osman Alperen Koraş, Meijie Li, Jörg Schlötterer 외

Pre-trained Language Models (PLMs) are trained on vast unlabeled data, rich in world knowledge. This fact has sparked the interest of the community in quantifying the amount of factual knowledge present in PLMs, as this …

Knowledge ProbingWorld Knowledge

UCoder: Unsupervised Code Generation by Internal Probing of Large Language Models

2025-12-19 · Jiajun Wu, Jian Yang, Wei Zhang, Lin Jing 외 arxiv

Large language models (LLMs) have demonstrated remarkable capabilities in code generation tasks. However, their effectiveness heavily relies on supervised training with extensive labeled (e.g., question-answering pairs) …

Code Generation