Search papers, labs, and topics across Lattice.
1
0
2
4
Turns out, ensembling LLMs for diversity isn't about picking the biggest model, but routing each prompt to the *right* model.
