Search papers, labs, and topics across Lattice.
1
0
2
LLMs can be made more reliable and efficient by adaptively focusing uncertainty quantification on relevant semantic themes, cutting inference time by 60% while improving factuality correlation.
