GitHub·Code·Apache-2.0

NVlabs/COAT

[ICLR 2025] COAT: Compressing Optimizer States and Activation for Memory-Efficient FP8 Training

263 stars·Updated Jul 2026
Open on GitHub

Linked papers