We like to invite you to our upcoming Invited talk by Dr. Wei-Fan Chen from University of Bonn. The talk will take place on Thursday, December 5, 2024 at 16:00, in Room 1101.F435 (Welfengarten 1):
# Speaker
Dr. Wei-Fan Chen
University of Bonn
https://caisa-lab.github.io/members/wei_fan.html
# Time and location
Thursday, December 5, 16:00
1101.F435 (Welfengarten 1)
# Title
Reasoning Hallucination in LLMs' Generated Texts
# Abstract
Hallucinations have been a critical issue in the generated texts by large language models (LLMs), which means LLMs produce unreliable and untrustable outputs. The hallucinated texts may sound plausible and fluent but contain incorrect or inconsistent information. In this talk, I will begin by introducing the types of hallucinations in LLMs, especially factualness and faithfulness. Following this, I will explore some popular methods and datasets in this research topic to detect or reduce them. Finally, I will present our ongoing research in detecting and reasoning hallucinations, where our goal is to provide LLM users with a human-understandable way to identify hallucinations in texts.
External participants (i.e. neither students nor employees of LUH) are kindly requested to send a short, informal email to office@ai.uni-hannover.de to register.
