GatedDeltaNet-2
Python
★ 221
updated 26d ago
Official PyTorch Implementation of Gated DeltaNet-2: Decoupling Erase and Write in Linear Attention
No plain-English explanation yet — one is being written right now. Check back in a minute.