EctoSpace / SCTView on GitHub
Spectral Compact Training (SCT): A methodology for extreme memory reduction in neural network training via dynamic SVD (W=Udiag(s)V^T) and Stiefel manifold retraction. Enables training large models in highly constrained hardware environments.
24May 26, 2026Updated last month

Alternatives and similar repositories for SCT

Users that are interested in SCT are comparing it to the libraries listed below. We may earn a commission when you buy through links labeled 'Ad' on this page.

Sorting:

Are these results useful?