Training language models on the knowledge graph: Insights on hallucinations and their detectability
TL;DR
Larger and longer-trained LMs hallucinate less on fixed data, but their remaining hallucinations become harder to detect.
Venue
First Conference on Language Modeling
BibTeX
@article{hron2024traininglanguagemodelsknowledge,
title={Training Language Models on the Knowledge Graph: Insights on Hallucinations and Their Detectability},
author={Jiri Hron and Laura Culp and Gamaleldin Elsayed and Rosanne Liu and Ben Adlam and Maxwell Bileschi and Bernd Bohnet and JD Co-Reyes and Noah Fiedel and C. Daniel Freeman and Izzeddin Gur and Kathleen Kenealy and Jaehoon Lee and Peter J. Liu and Gaurav Mishra and Igor Mordatch and Azade Nova and Roman Novak and Aaron Parisi and Jeffrey Pennington and Alex Rizkowsky and Isabelle Simpson and Hanie Sedghi and Jascha Sohl-dickstein and Kevin Swersky and Sharad Vikram and Tris Warkentin and Lechao Xiao and Kelvin Xu and Jasper Snoek and Simon Kornblith},
year={2024},
eprint={2408.07852},
archivePrefix={arXiv},
primaryClass={cs.CL}
title={Training Language Models on the Knowledge Graph: Insights on Hallucinations and Their Detectability},
author={Jiri Hron and Laura Culp and Gamaleldin Elsayed and Rosanne Liu and Ben Adlam and Maxwell Bileschi and Bernd Bohnet and JD Co-Reyes and Noah Fiedel and C. Daniel Freeman and Izzeddin Gur and Kathleen Kenealy and Jaehoon Lee and Peter J. Liu and Gaurav Mishra and Igor Mordatch and Azade Nova and Roman Novak and Aaron Parisi and Jeffrey Pennington and Alex Rizkowsky and Isabelle Simpson and Hanie Sedghi and Jascha Sohl-dickstein and Kevin Swersky and Sharad Vikram and Tris Warkentin and Lechao Xiao and Kelvin Xu and Jasper Snoek and Simon Kornblith},
year={2024},
eprint={2408.07852},
archivePrefix={arXiv},
primaryClass={cs.CL}
Date
August, 2024
Links