Search papers, labs, and topics across Lattice.
1
0
2
Stop wasting compute on LLM evals: a variance-adaptive querying strategy slashes estimation error by focusing on the most uncertain prompt-response pairs.
