Search papers, labs, and topics across Lattice.
Affiliation:
2
0
3
0
Cascaded batch prompting resolves the unpredictability of conventional batch prompting, achieving both superior performance and efficiency in large language models.
Self-consistency, previously known for boosting reasoning, turns out to be a surprisingly effective way to make LLMs remember facts too.
