Search papers, labs, and topics across Lattice.
1
0
2
Consolidating specialist LLMs into a single, more accurate generalist is possible with *zero* gradient updates, relying solely on forward passes and input statistics.