Search papers, labs, and topics across Lattice.
1
0
3
2
Reasoning-heavy LLMs can cut their "thinking" time by up to 72% with a new scheduling algorithm that understands when the model is reasoning vs. answering.
