Search papers, labs, and topics across Lattice.
3
0
4
Imperfect LLM detectors can paradoxically drive users to increase their reliance on LLMs, ultimately degrading output quality.
Excluding features based on manipulability can lead to suboptimal predictions, revealing a critical flaw in standard feature selection practices.
Aggregating responses from multiple copies of the same model expands the range of achievable outputs in compound AI systems through three key mechanisms, offering a path to overcome individual model limitations.
