news

Oct 10, 2024 Our paper “Are LLMs Prescient? A Continuous Evaluation using Daily News as the Oracle” is accepted in NeurIPS Workshop on Adaptive Foundation Models (Oral)!
Jun 01, 2024 We finished our project “DENIAHL: In-Context Features Influence LLM Needle-In-A-Haystack Abilities”, where we analyze factors beyond context length affecting LLMs’ abilities to recall information from long input context. Check our paper and code.
May 05, 2024 We revisited the Text-Based Ideal Point Model (TBIP)! TBIP (Vafa et al., 2020) is a probabilistic topic model for estimating ideologies using word-choice differences on shared topics, we explored its potential in multiparty contexts! See our paper and code.