5 papers
What do Geometric Hallucination Detection Metrics Actually Measure?
Eric Yeats, John Buckheit, Sarah Scullen +9
Hallucination remains a barrier to deploying generative models in high-consequence applications. This is especially true in cases where external ground truth is not readily availab…
Saddle-Free Guidance: Improved On-Manifold Sampling without Labels or Additional Training
Eric Yeats, Darryl Hannan, Wilson Fearn +3
Score-based generative models require guidance in order to generate plausible, on-manifold samples. The most popular guidance method, Classifier-Free Guidance (CFG), is only applic…
A Connection Between Score Matching and Local Intrinsic Dimension
Eric Yeats, Aaron Jacobson, Darryl Hannan +4
The local intrinsic dimension (LID) of data is a fundamental quantity in signal processing and learning theory, but quantifying the LID of high-dimensional, complex data has been a…
Automating Evaluation of Diffusion Model Unlearning with (Vision-) Language Model World Knowledge
Eric Yeats, Darryl Hannan, Henry Kvinge +2
Machine unlearning (MU) is a promising cost-effective method to cleanse undesired information (generated concepts, biases, or patterns) from foundational diffusion models. While MU…
Min-K%++: Improved Baseline for Detecting Pre-Training Data from Large Language Models
Jingyang Zhang, Jingwei Sun, Eric Yeats +5
The problem of pre-training data detection for large language models (LLMs) has received growing attention due to its implications in critical issues like copyright violation and t…