Adaptive Activation Steering: A Tuning-Free LLM Truthfulness Improvement Method for Diverse Hallucinations Categories
Top Cited Papers
Open Access
- 22 April 2025
- conference paper
- Published by Association for Computing Machinery (ACM)
- p. 2562-2578
Abstract
No abstract availableAll Related Versions
Funding Information
- Beijing Natural Science Foundation (Nos. L244063)
- Peking University Medicine plus X Pilot Program-Key Technologies R&D Project (2024YXXLHGG007)
- Health Data Research UK-The Alan Turing Institute Wellcome PhD Programme in Health Data Science (Grant Ref: 218529/Z/19/Z)
- National Natural Science Foundation of China (62402017, U23A20468)
- Xuzhou Scientific Technological Projects (KC23143)
This publication has 6 references indexed in Scilit:
- Explainable Fake News Detection with Large Language Model via Defense Among Competing Wisdom2024
- Towards Trustworthy Large Language Models2024
- Understanding the role of individual units in a deep neural networkProceedings of the National Academy of Sciences, 2020
- Interpreting the Latent Space of GANs for Semantic Face Editing2020
- Natural Questions: A Benchmark for Question Answering ResearchTransactions of the Association for Computational Linguistics, 2019
- Deep Feature Interpolation for Image Content Changes2017