Reproduce “Continual Learning in Modern Hopfield Networks with an Application to Diffusion Models” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Ken Takeda, Masafumi Oizumi, Ryo Karakida
The authors test whether Hopfield energy identifies old images vulnerable to forgetting during sequential diffusion-model fine-tuning. In split CIFAR-10, higher Task-1 energy is associated with higher post-Task-2 reconstruction mean squared error in Stable Diffusion v1.5 and a pixel-space DDPM. High-energy replay gives concentrated rescue, but reconstruction error is only a proxy for forgetting.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.