Search papers, labs, and topics across Lattice.
1
0
3
9
Forget scaling up: smarter state mixing in linear recurrent networks lets you match LSTMs and Mamba on sequence tasks, closing the expressivity gap without sacrificing efficiency.
