Learning to poison large language models during instruction tuning

Y Qiang, X Zhou, SZ Zade, MA Roshani… - ar** public opinion. Despite journalism's aim for
impartial reporting, various biases can emerge during writing and publication phases. While …

[HTML][HTML] Harnessing Response Consistency for Superior LLM Performance: The Promise and Peril of Answer-Augmented Prompting

H Wu, H Hong, L Sun, X Bai, M Pu - Electronics, 2024 - mdpi.com
This paper introduces Answer-Augmented Prompting (AAP), an innovative approach that
leverages the Response Consistency of History of Dialogue (HoD) phenomenon in Large …

Intended Target Identification for Anomia Patients with Gradient-based Selective Augmentation

J Kim, R Storaï, S Hwang - Findings of the Association for …, 2024 - aclanthology.org
In this study, we investigate the potential of language models (LMs) in aiding patients
experiencing anomia, a difficulty identifying the names of items. Identifying the intended …

Hijacking Large Language Models via Adversarial In-Context Learning

Y Qiang - 2024 - search.proquest.com
In-context learning (ICL) has emerged as a powerful paradigm leveraging LLMs for specific
downstream tasks by utilizing labeled examples as demonstrations in the precondition …

[PDF][PDF] On Learning Frequency-Instance Correlations by Model-Agnostic Training for Synthetic Speech Detection

Z Wang, L Gao, J Zhang, Q Mao - The 16th Asian …, 2024 - raw.githubusercontent.com
Abstract The goal of Synthetic Speech Detection (SSD) is to detect spoofing speech
synthesized by text-to-speech and voice conversion. Most existing SSD methods focus only …