Showing 2025 · cs.CLShow all
2 papers · 2 filters
cs.CL2025
Probing the Geometry of Truth: Consistency and Generalization of Truth Directions in LLMs Across Logical Transformations and Question Answering Tasks
Yuntai Bao, Xuhong Zhang, Tianyu Du +4
Large language models (LLMs) are trained on extensive datasets that encapsulate substantial world knowledge. However, their outputs often include confidently stated inaccuracies. E…
cs.CL2025
Scalable Multi-Stage Influence Function for Large Language Models via Eigenvalue-Corrected Kronecker-Factored Parameterization
Yuntai Bao, Xuhong Zhang, Tianyu Du +4
Pre-trained large language models (LLMs) are commonly fine-tuned to adapt to downstream tasks. Since the majority of knowledge is acquired during pre-training, attributing the pred…