Search papers, labs, and topics across Lattice.
1
0
3
Salience Bias in LLMs reveals that models often ignore commonsense reasoning in favor of misleading explicit cues, with lightweight prompting showing promise in addressing this issue.
