Search papers, labs, and topics across Lattice.
1
0
2
Nearly half of published findings about LLM behavior in conversations may be statistically invalid due to a widespread failure to account for autocorrelation between turns.