no code implementations • 17 Dec 2023 • Jiachuan Wang, Shimin Di, Lei Chen, Charles Wang Wai Ng
We validate our conjecture that monosemanticity brings about performance change at different model scales on a variety of neural networks and benchmark datasets in different areas, including language, image, and physics simulation tasks.
1 code implementation • ICCV 2023 • Jiachuan Wang, Shimin Di, Lei Chen, Charles Wang Wai Ng
However, such a method is highly sensitive to the standard deviation \sigma_n of noises injected to clean images, where \sigma_n is inaccessible without knowing clean images.