Search papers, labs, and topics across Lattice.
1
0
3
Fast weight models can now achieve significantly better long-context performance thanks to a new RL training framework that moves beyond next-token prediction.
