Search papers, labs, and topics across Lattice.
1
0
2
3
Scaling up LLMs boosts combinatorial creativity, but hits diminishing returns for exploratory tasks, revealing a "convergence-by-scaling" effect where larger models become less divergent.
