Reproduce “Event-triggered Implicit Perturbation for Zeroth-Order Fine-Tuning of Spiking Transformers” 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
Tengteng Lei, Prabodh Katti, Rashi Dutt, Houssem Sifaou, Tan Peng, Osvaldo Simeone, Kai Xu, Bipin Rajendran
IPZO enables edge adaptation of spiking neural networks through zeroth-order learning. Rather than rewriting in-memory computing weights for each perturbation, it computes the perturbation contribution separately and adds it to the output. On CIFAR-10, PGU-XOR reaches 76.41% accuracy, versus 66.85% for PGU-Reuse, reaching their shared target in 84 rather than 247 epochs. Modeled energy is 0.83× EPZO at B=64, T=4, not total accelerator energy.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.