Search papers, labs, and topics across Lattice.
1
0
3
Achieving a 6.47x decoding speedup while improving accuracy on long-context tasks could redefine efficiency benchmarks in language modeling.
