Search papers, labs, and topics across Lattice.
2
0
3
2
Exploit the surprisingly stable, yet heterogeneous, sparsity patterns across attention heads to slash LLM attention latency by 2.88x without sacrificing quality.
Beat the LLM inference bottleneck: SageSched's uncertainty-aware scheduling boosts efficiency by nearly 30% by predicting output length and balancing compute and memory demands.
