Seminars
LLMs in Recommender System
Wenwei Ban
June 27, 2025
This talk is about LLMs in RecSys, using LLMs to enhance the performance of recommender system.
View PDFReLearn: Unlearning via Learning for Large Language Models
Honglin Wang
June 13, 2025
This talk is about ReLearn, a novel unlearning framework that uses positive optimization to balance a model's forgetting, retention, and linguistic capabilities.
View PDFAI Scheming: Frontier Models may Pursue Secret Goals and Lie to You
Xun Liu
May 23, 2025
This talk introduces the concept of AI scheming and discusses two paper under this topic, with the focus on in-context scheming and harmfulness sandbagging for each.
View PDFRethinking Self-Consistency for Hallucination Detection in LLMs
Sha Liu
May 09, 2025
This talk discusses the self-consistency methods in hallucination detection.
View PDFTowards Robust Safety Alignment of LLMs
Jiajun Tan
April 11, 2025
This talk discusses two paper about issues LLM safety alignment, mainly focusing on vulnerabilities against attacks and approaches to improve the robustness of alignment.
View PDFRethinking LLM Unlearning: Benchmarks and Datasets
Ruichen Qiu
March 21, 2025
This talk discusses two work about LLM unlearning, each focusing on a specific issue in the field and proposing a new benchmark for better evaluation.
View PDFReasoning Errors of LLMs
Wanli Yang
March 14, 2025
This talk focses on the self-correction capabilities of LLMs in reasoning, revealing existing LLMs cannot self-correct reasoning yet.
View PDF