Search papers, labs, and topics across Lattice.
1
0
3
2
LLM win rates in multi-agent games can nearly double (from 25% to 50%) simply by optimizing the context provided during inference.
